Target detection method, and integrated circuit, sensor, device and medium

Abstract

A target detection method, and an integrated circuit, a sensor, a device and a medium. The method comprises: performing frame data accumulation on 1D-FFT data, wherein the 1D-FFT data is obtained by means of performing distance-dimension FFT processing on an echo signal; performing inter-frame FFT processing on the basis of the accumulated frame data, so as to obtain an RD map; and detecting a target in a region of interest on the basis of the RD map.

Description

Target detection method, integrated circuit, sensor, device and medium

[0001] This application claims priority to the Chinese patent application filed on July 24, 2023, with application number 202310913653.8 and invention name “Target Detection Method and System, Integrated Circuit, Sensor and Device”, the contents of which should be understood as incorporated into this application by reference. Technical Field

[0002] The embodiments of the present disclosure relate to, but are not limited to, the field of electromagnetic wave sensor technology, and specifically to a target detection method and system, an integrated circuit, a sensor, and a device. Background Art

[0003] When performing target detection in sealed or relatively sealed spatial areas, such as target detection in a car cabin, there may be technical problems such as low detection rate, a large number of false alarm targets, and inaccurate angle estimation.

[0004] Summary of the Invention

[0005] The following is a summary of the subject matter described in detail herein. This summary is not intended to limit the scope of the claims.

[0006] An embodiment of the present application provides a method for target detection, which may include:

[0007] accumulating frame data of 1D-FFT data, where the 1D-FFT data is obtained by performing range-dimensional FFT processing on the echo signal;

[0008] Performing inter-frame FFT processing based on the accumulated frame data to obtain an RD spectrum; and

[0009] Detection of targets in the region of interest is achieved based on the RD map.

[0010] Optionally, the method may include: accumulating frame data of the 1D-FFT data until a preset amount of data is accumulated, and then performing FFT processing between frames to obtain an RD spectrum.

[0011] Optionally, for multi-channel application scenarios, the detection of targets in the area of ​​interest based on the RD spectrum may include: after incoherent accumulation of the RD spectrum of each channel, continuing with constant false alarm processing based on noise estimation, wave direction estimation, and target judgment, positioning and recognition operations.

[0012] Optionally, the constant false alarm processing may be performing DAE CFAR processing on the azimuth angle and the elevation angle respectively.

[0013] Optionally, the detection of the target in the region of interest may include: performing a target recognition operation based on extracting at least part of the multi-frame accumulated data, the inter-frame Fourier transform data, the incoherent accumulation data and / or the constant false alarm detection data.

[0014] Optionally, for multi-channel application scenarios, the detection of targets in the area of ​​interest based on the RD spectrum may include: performing constant false alarm processing on the RD spectrum of each channel respectively, continuing with binary integration processing in the channel domain and wave direction estimation, and then performing target judgment, positioning and recognition operations.

[0015] Optionally, the detection of the target in the region of interest based on the RD map includes: performing two-dimensional digital beam synthesis on the RD map, and continuing basic target classification processing to achieve target judgment, positioning and recognition operations.

[0016] Optionally, performing deep learning target detection, deep learning target positioning and / or deep learning target recognition on the 1D-FFT data may include: obtaining target point cloud data based on the 1D-FFT data; and performing ML-based false alarm suppression, clustering processing and / or ML-based target classification processing on the target point cloud data to achieve judgment, positioning and / or recognition operations on targets in the area of ​​interest.

[0017] Optionally, the method may also include: performing two-dimensional digital beam synthesis on the 1D-FFT data, continuing to accumulate frame data until a preset amount of data is accumulated, and performing inter-frame FFT processing to obtain an RD spectrum; and performing DL-based target classification processing based on the RD spectrum to achieve target judgment, positioning and recognition operations.

[0018] An embodiment of the present application also provides a method for target detection, which may include: performing distance-dimensional FFT processing based on the echo signal to obtain 1D-FFT data; performing two-dimensional digital beam synthesis on the 1D-FFT data, and continuing to accumulate frame data; and when a preset amount of data is accumulated, performing DL-based target classification processing to achieve target judgment, positioning and identification operations.

[0019] An embodiment of the present application also provides an integrated circuit, which may include: a signal transmitting module, configured to transmit electromagnetic waves for target detection; a signal receiving module, configured to receive echoes formed by the reflection and / or scattering of the electromagnetic waves; and a processing module, configured to perform signal and data processing on the echoes according to any of the methods of the aforementioned embodiments to achieve target detection.

[0020] Optionally, the processing module includes a baseband unit and an MCU unit. The baseband unit can be configured to implement the distance-dimensional FFT processing and inter-frame FFT processing in the method described in any embodiment of the present application. The MCU unit can be configured to implement the frame data accumulation and detection of targets in the area of ​​interest in the method described in any embodiment of the present application.

[0021] Optionally, when the method includes digital beamforming and frame data accumulation, the baseband unit may be configured to implement the digital beamforming, and the MCU unit may be configured to implement the frame data accumulation.

[0022] Optionally, the integrated circuit is a millimeter wave chip or a sensor chip.

[0023] An embodiment of the present application also provides an electromagnetic wave sensor, which may include: a carrier; an integrated circuit as described in any embodiment of the present application, which is arranged on the carrier; an antenna, which is arranged on the carrier, or the antenna and the integrated circuit are integrated into a single device and arranged on the carrier; wherein the integrated circuit is connected to the antenna for transmitting the electromagnetic wave signal and / or receiving the echo signal.

[0024] An embodiment of the present application also provides a terminal device, which may include: a device body; and an electromagnetic wave sensor as described in any embodiment, which is arranged on the device body; wherein the electromagnetic wave sensor is used for target detection and / or communication to provide reference information for the operation of the device body.

[0025] An embodiment of the present application also provides a non-transitory computer-readable storage medium on which computer-readable instructions are stored. When the instructions are executed by a processor, the processor executes the method described in any embodiment of the present application.

[0026] Still other aspects will become apparent upon reading and understanding the accompanying drawings and detailed description.

[0027] Summary of the Figures

[0028] The above and other objects, features and advantages of the present application will become more apparent through the following description of the embodiments of the present application with reference to the accompanying drawings.

[0029] FIG1 is a schematic diagram of a process for implementing target detection based on SISO-combine in an embodiment of the present application;

[0030] FIG2 is a schematic diagram of another process for implementing target detection based on SISO-combine in an embodiment of the present application;

[0031] FIG3 is a flow chart of a target detection method provided in an embodiment of the present application;

[0032] FIG4 is a schematic diagram of a working cycle of a radar provided in an embodiment of the present application;

[0033] FIG5 is a schematic diagram of a waveform of a detection signal of an FMCW radar provided in an embodiment of the present application;

[0034] FIG6 is another flow chart of the target detection method provided in an embodiment of the present application;

[0035] FIG7 is another flow chart of the target detection method provided in an embodiment of the present application;

[0036] FIG8 is another flow chart of the target detection method provided in an embodiment of the present application;

[0037] FIG9 is another flow chart of the target detection method provided in an embodiment of the present application;

[0038] FIG10 is a schematic diagram of data reading involved in the target detection method provided in an embodiment of the present application;

[0039] FIG11 is another flow chart of the target detection method provided in an embodiment of the present application;

[0040] FIG12 is another flow chart of the target detection method provided in an embodiment of the present application;

[0041] FIG13 is a schematic diagram of data movement involved in the target detection method provided in an embodiment of the present application;

[0042] FIG14 is another flow chart of the target detection method provided in an embodiment of the present application;

[0043] FIG15 is a schematic diagram of the division of a preset area involved in the target detection method provided in an embodiment of the present application;

[0044] FIG16 is another flow chart of the target detection method provided in an embodiment of the present application;

[0045] FIG17 is another flow chart of the target detection method provided in an embodiment of the present application;

[0046] FIG18 is a schematic diagram showing the distribution of target points in a preset area involved in the target detection method provided in an embodiment of the present application;

[0047] FIG19 is another flow chart of the target detection method provided in an embodiment of the present application;

[0048] FIG20 is a schematic diagram of a process for implementing target detection based on per-channel in an embodiment of the present application;

[0049] FIG21 is a flow chart of a target detection method provided in an embodiment of the present application;

[0050] FIG22 is another flow chart of the target detection method provided in an embodiment of the present application;

[0051] FIG23 is another flow chart of the target detection method provided in an embodiment of the present application;

[0052] FIG24 is a flowchart illustrating target detection after combining the target detection method provided in an embodiment of the present application with the multi-frame joint processing solution;

[0053] FIG25 is another flow chart of the target detection method provided in an embodiment of the present application;

[0054] FIG26 is another flow chart of the target detection method provided in an embodiment of the present application;

[0055] FIG27 is another flow chart of the target detection method provided in an embodiment of the present application;

[0056] FIG28 is a schematic diagram of an output involved in the target detection method provided in an embodiment of the present application;

[0057] FIG29 is a flow chart illustrating target detection based on frame-level FFT combined with DAE CFAR in an embodiment of the present application;

[0058] FIG30 is a schematic diagram of azimuth or elevation CFAR in the target detection process shown in FIG3 ;

[0059] FIG31 is a flow chart of a target detection method according to an embodiment of the present disclosure;

[0060] FIG32 is a flow chart illustrating target detection based on frame-level FFT combined with DAE CFAR in an embodiment of the present disclosure;

[0061] FIG33 is a schematic diagram of the result after incoherent accumulation;

[0062] Figure 34 is a partial diagram of the incoherent accumulation result;

[0063] FIG35 is a schematic diagram of target points in the sixth area according to an embodiment of the present disclosure;

[0064] Figures 36A-36D show the processing results for a single-person scenario (baby in aisle C);

[0065] 37A-37D show the processing results for a two-person scenario (the baby is in seat B and the adult is in seat A).

[0066] FIG38 is a schematic diagram of a process for post-processing a target in an embodiment of the present application;

[0067] FIG39 is a schematic diagram of a process for post-processing a target in combination with a DL algorithm in an embodiment of the present application;

[0068] FIG40 is a flow chart of a target detection method incorporating a DL algorithm in an embodiment of the present application;

[0069] FIG41 is a flow chart of another target detection method combined with a DL algorithm in an embodiment of the present application;

[0070] Figure 42 is a flow chart of another target detection method combined with the DL algorithm in an embodiment of the present application.

[0071] Details

[0072] To facilitate understanding of the present application, a more comprehensive description of the present application will be provided below with reference to the accompanying drawings. The accompanying drawings illustrate preferred embodiments of the present application. However, the present application may be implemented in various forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to facilitate a more thorough and comprehensive understanding of the present application.

[0073] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application pertains. The terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit this application.

[0074] The present disclosure provides a method for detecting an object, including:

[0075] S1, accumulating frame data of 1D-FFT data, wherein the 1D-FFT data is obtained by performing distance-dimensional FFT processing on the echo signal;

[0076] S2, performing inter-frame FFT processing based on the accumulated frame data to obtain an RD spectrum; and

[0077] S3, detecting the target in the region of interest based on the RD map.

[0078] Exemplarily, all of the 1D-FFT data after 1D-FFT processing can be accumulated between frames for inter-frame FFT processing, or part of the 1D-FFT data can be accumulated between frames. For example, when the region of interest is smaller than the range of the echo data, only part of the echo signal can be processed by distance-dimensional FFT to obtain 1D FFT data, and then subsequent inter-frame accumulation and inter-frame FFT processing can be performed, or part of the 1D-FFT data can be accumulated between frames and processed by inter-frame FFT.

[0079] Exemplarily, the range-dimensional FFT processing and the inter-frame FFT processing may be performed simultaneously. For example, while the inter-frame FFT processing is being performed, the range-dimensional FFT processing continues to be performed.

[0080] When detecting targets in sealed or relatively sealed spatial areas, in some optional embodiments, target detection can be achieved by performing regional judgment based on single-frame fast and slow time processing, CFAR (Constant False-Alarm Rate) detection, and angle measurement to obtain a point cloud. Alternatively, single-frame fast time processing can be followed by performing DoA (Direction of Arrival) estimation using algorithms such as Capon to form an RA (Range-Azimuth) map, and then extracting features such as energy within a predetermined area for identification and judgment. Alternatively, target detection can be achieved by filtering the received echo phase and then detecting human respiration and heartbeat. For example, when target detection is achieved by performing regional judgment based on single-frame fast and slow time processing, CFAR detection, and angle measurement to obtain a point cloud, 1D-FFT (e.g., range-dimensional FFT) processing can be performed on the echo signal, and then time correlation analysis can be performed based on the complex echo data to achieve human target identification. Subsequently, position judgment can be performed by combining point cloud information to ultimately achieve the purpose of in-cabin target detection. For example, when target detection is achieved by performing regional judgment based on the point cloud obtained by single-frame fast and slow time processing, CFAR detection, and angle measurement, the echo signal can be processed by 1D-FFT (such as range-dimensional FFT) and then time correlation analysis can be performed based on the complex echo data to achieve human target discrimination. Subsequently, position judgment can be performed by combining point cloud information, etc., to ultimately achieve the purpose of in-cabin target detection.

[0081] In some optional embodiments of the present application, another target detection and / or recognition scheme is also proposed, that is, a scheme for realizing target detection in sealed space areas such as cabins, indoors, and factories through inter-frame accumulation, so as to effectively improve the detection rate, reduce the number of false alarm targets, and significantly improve the accuracy of angle estimation, etc., so that special targets or weak targets such as infants and young children in the cabin can be accurately detected, realizing applications such as CPD (Child Presence Detection) and SBR (Safety Belt Reminder). In the process of implementation, accurate detection of targets in the cabin can be achieved based on a combination of one or at least two schemes such as multi-frame joint processing technology and deep learning detection, positioning and recognition technology.

[0082] The multi-frame joint processing solution can be to obtain a range-Doppler map by performing a sliding window FFT on the multi-frame data, or it can be processed by using FIR (Finite Impulse Response) or other complex time-frequency transforms, and then perform CFAR, DOA (Direction of Arrival) and other processing operations on the area of ​​interest. It can also be combined with the set area judgment logic and area parameters to achieve the detection and positioning of living targets such as adults, children, and pets, or other non-living targets. CFAR can be NR-CFAR in the Doppler dimension, or RD-CFAR, or DAE (Doppler-Azimuth-Elevation)-CFAR, etc. At the same time, CFAR and DoA can be followed by post-processing such as clustering, false alarm suppression, and multi-processing point cloud association. The regional parameters can be determined by adopting at least one of the following schemes, or a combination of at least two schemes, such as clustering operation on the point cloud detected after a certain sliding window multi-frame processing, outlier detection and elimination operation on the point cloud detected after a certain sliding window multi-frame processing, etc.

[0083] When applied to enclosed or relatively enclosed spaces, the system can be pre-divided to define and detect different areas of interest (divided areas). For example, when detecting targets within a vehicle cabin, the radar monitoring area can be divided into seating areas and aisle areas, for example. Parameter types and thresholds can then be preset for each type of area, and the corresponding processing steps can be combined to achieve precise detection of targets of interest within a specific area.

[0084] In some optional embodiments, for family cars, the interior space can generally be simply divided into headroom, rear space, and trunk space. If the rear space is the key monitoring area, it can be further divided into seat zones and aisle zones. The seat zones can also be divided into a corresponding number of seat zones based on the number of seats. Similarly, the aisle zones can be divided into a corresponding number of aisle zones based on the number of seats. For example, for a five-seat family sedan, the three seats in the rear space can be divided into three seat zones and three aisle zones. Furthermore, corresponding parameter types and thresholds can be preset for different types of zones (or zones or zones), and adaptive signal data processing methods and steps can be employed to achieve accurate detection of targets of interest (specific targets) within specific zones or zones. Adjacent zones can have partial overlap or be adjacent or separated by a gap of a preset width.

[0085] When combining deep learning with detection, positioning, and identification, the implementation can be as follows: performing 1D-FFT (e.g., range-dimensional FFT) processing on the echo signal, followed by a multi-frame sliding window FFT, and then performing N-point 2D-DBF processing to form an N-channel RD (Range-Doppler) map, and then using a deep learning model to perform monitoring, positioning, and identification operations based on the N-channel RD map; or performing 1D-FFT processing on the echo signal, followed by N-point 2D-DBF processing, and then using a deep learning model to perform monitoring, positioning, and identification operations based on the N-channel RD map; or performing 1D-FFT processing on the echo signal, followed by N-point 2D-DBF processing, and then performing multi-frame sliding window FFT to form an N-channel RD map, and then using a deep learning model to perform monitoring, positioning, and identification operations based on the N-channel RD map. The aforementioned N can be the number of interval units; for example, for a rear row with three seats, the rear row is divided into three seat interval units and three aisle interval units, and N can be equal to 6.

[0086] Based on the above solution, after testing using real-world data, it was shown that for top-mounted radar installation scenarios, a high detection rate, as well as extremely low missed detection and false alarm rates, can be achieved. For example, when detecting targets within a cabin, a detection rate of over 99% can be achieved, a missed detection rate of less than 0.5%, and a false alarm rate of less than 1%.

[0087] The following uses frame-level Fourier transform as an example and describes the technical solution of the present application in detail with reference to the accompanying drawings. That is, after performing a fast-time Fourier transform on the chirp in the echo signal, at least two frames of fast-time Fourier transform data are accumulated, and then a frame-dimensional Fourier transform is performed on the at least two frames of data to obtain a range-Doppler (RD) spectrum, so as to estimate the target distance and / or velocity based on the RD spectrum.

[0088] FIG1 is a flow chart of target detection based on SISO-combine in an embodiment of the present application. As shown in FIG1 , after performing operations such as ADC (analog-to-digital) conversion and sampling processing on the echo signal, range DC removal and range Fourier transform (also expressed as 1D-FFT or fast time FFT) are sequentially performed, followed by multi-frame accumulation (e.g., storing 128 frames of data), Doppler DC removal, frame FFT, and non-coherent integration. Constant false alarm detection (e.g., peak-based CFAR (detection with peak selection)) is then performed based on the noise obtained by noise estimation (e.g., using a noise variance estimator). Finally, angle detection and target classification are performed to obtain target range, velocity, and / or angle information. For example, azimuth and elevation estimation can be performed based on the CFAR results, for example, by using a digital beamformation (DBF) (DBF). It is implemented using technologies such as digital beamforming (forming) and DoA (direction of arrival estimation).

[0089] In some optional embodiments, for an electromagnetic wave sensor having a BB (Baseband) unit and an MCU (Microcontroller Unit) unit, based on the flowchart of target detection shown in Figure 1, operations such as frame data accumulation (store 128 frames) and target classification (Region HIST, classification) can be performed in the MCU module, while range DC removal (Range DC removal) and Doppler DC removal (Doppler DC removal) can be performed in the BB module or the MCU module. In the process shown in Figure 1, DC filtering of downlink ADC data in the BB module is taken as an example for illustration.

[0090] As shown in Figure 1, after performing 1D-FFT processing or obtaining distance dimension information in the BB module, the 1D-FFT data can be cached in the MCU module. When the cache reaches a preset data amount (such as 128, 56, or 32 frames), the data cached to the preset number of frames can be subjected to DC removal and frame-level FFT in the frame dimension to obtain an RD (Range-Doppler) diagram.

[0091] When the system includes multiple channels, as shown in Figure 1, non-coherent integration can be performed on the RD spectra of the multiple channels, and the noise floor of each range bin in the accumulated data can be estimated using the NVE module. For example, the noise floor estimation can be performed based on a preset formula, and subsequent NR (Noise Reference)-CFAR processing can be performed based on the value of the noise floor estimation. The above preset formula can be: i =min(n i ,α*n g ), n i is the noise floor estimate of the i-th range bin, n g is the noise floor estimate of the last range bin of interest, and α is a coefficient that can be set and updated based on requirements, engineering data, experience, etc.

[0092] For angle estimation, as shown in Figure 1, DBF and DoA in the azimuth dimension, as well as DBF and DoA in the elevation dimension, can be performed on the CFAR detection point data to obtain azimuth and elevation angle estimates for each CFAR detection point. Furthermore, the number of target points detected in each frame (or a preset amount of data, i.e., each time) can be counted according to a preset (or divided) regional unit design. The count results can be judged according to preset rules to determine whether a physical target of interest (such as an adult, child, infant, or pet) is present in this processing and determine its location area.

[0093] In the above-mentioned embodiment, the target's speed is determined by using a frame-level FFT instead of the traditional chirp-level Doppler FFT to achieve accurate detection of the target of interest, and the frequency of updating the results can be further increased by performing sliding window FFT processing on multiple frames, thereby improving the real-time performance of the system. At the same time, the amount of data to be processed can be reduced by processing only the distance and / or Doppler area of ​​interest. In addition, when estimating the background noise, the background noise is obtained by taking the minimum value based on the estimate of the current range bin combined with the estimate of the last range bin of interest, which can be more suitable for application scenarios in relatively sealed environments such as the interior of a car. When judging the regional logic, by counting the target points detected in each frame according to the preset area design and judging the counting results according to the preset rules, it is possible to accurately judge whether there is a target in each preset area.

[0094] Figure 2 shows an example of digital signal processing implemented by another target detection method applied to a SISO-combine radar.

[0095] After receiving the echo signal through the antenna and performing analog-to-digital conversion, sampling, and other processing, the radar obtains a digital signal. This digital signal is then processed by a main control module (e.g., which can be implemented based on a microcontroller unit (MCU)) and a baseband (BB) module (e.g., which can be implemented based on a baseband chip) as shown in Figure 2 (taking target detection of a frequency modulated continuous wave (FMCW) radar as an example) to achieve target detection. For example, in Figure 2, the digital signal undergoes range DC removal and range Fourier transform (Range FFT, also expressed as 1D-FFT or fast time FFT) on the baseband module. The baseband module can then further process the data, for example, combining multiple chirp data in a frame using complex averaging. Alternatively, no further processing is required. Afterwards, the signal processed by the baseband module is sent to the main control module for multi-frame accumulation. When the data accumulates to a preset number of frames, such as 128 frames, the main control module transfers the accumulated data from the CPU static random-access memory (SRAM) to the baseband module. The baseband module then performs Doppler DC removal, frame Fourier transform (FFT), and inter-channel accumulation (non-coherent integration processing, channel accumulation, etc.). The baseband module then performs constant false alarm detection (e.g., peak-based CFAR (detection with peak selection)) based on the noise obtained by noise estimation (e.g., using a variance-based noise estimator). Finally, angle detection and target classification are performed to obtain target range, velocity, and / or angle information. When performing angle detection, azimuth and elevation estimation can be performed based on the CFAR results. This can be achieved, for example, through technologies such as digital beam forming (DBF) and direction of arrival estimation.

[0096] The implementation of the digital signal processing shown in FIG2 will be described below.

[0097] Range-dimensional DC removal targets the data corresponding to each chirp in each frame. This is accomplished by averaging the data collected by each chirp and each receive (RX) channel along the fast time dimension, then subtracting the DC component from all sampling points in each RX channel.

[0098] The distance dimension FFT can be implemented by windowing the data after the DC of the distance dimension is removed and then performing 1D-FFT.

[0099] It is important to note that during target detection, each chirp data in each frame of data obtained by radar through ADC or sampling based on the echo signal is processed as described above and sent to the main control module. This data can be sent to the main control module by moving the data from Direct Memory Access (DMA) to the CPU SRAM.

[0100] Doppler dimension DC removal can be achieved by calculating the complex average along the frame dimension of the multi-frame data accumulated on the main control module (that is, averaging the data of different frames in the same distance unit (bin) of the same channel), obtaining the average value of different distances in different channels as the DC component of the corresponding distance unit of the corresponding channel, and subtracting the respective DC components from each distance unit of each channel.

[0101] The inter-frame Fourier transform can be implemented by windowing the data after the Doppler dimension DC is removed along the frame dimension and performing 2D-FFT along the frame dimension. At this time, the range dimension-Doppler information of each transmitting and receiving channel will be obtained.

[0102] Inter-channel accumulation, taking incoherent accumulation as an example, can be implemented by the following expression:

[0103] or,

[0104] where P(r,v) and A(r,v) are the power and amplitude at the rth range unit and the vth Doppler unit, respectively, and s(r,t,a,v) is the complex value obtained by inter-frame Fourier transform of the signal echo of the rth range unit and the vth Doppler unit on the tth transmit channel and the ath receive channel.

[0105] The constant false alarm detection based on the noise floor obtained by noise estimation has been described in the existing CFAR technology and will not be described in detail here.

[0106] Angle detection can be achieved by performing digital beam forming (DBF) and DOA in azimuth, as well as DBF and DOA in elevation, on the target point obtained by CFAR. DBF and DOA have been described in existing DBF and DOA technologies and will not be elaborated on here.

[0107] Target classification can be achieved by counting the target points detected in each frame (or preset data amount, i.e. each time) according to a preset (or divided) area unit design, and the counting results can be judged according to preset rules to obtain whether there is a physical target of interest (such as an adult, child, infant or pet, etc.) in this processing and determine its location area.

[0108] It should be noted that FIG2 is merely an example for a radar having a baseband module and a main control module. That is, operations such as frame data accumulation and target classification (which can be implemented using Region HIST or classification) are performed in the main control module, while other operations, such as range-dimensional DC removal and Doppler-dimensional DC removal, are performed in the baseband module. In other embodiments, frame data accumulation may be performed in the baseband module, and Doppler-dimensional DC removal may be performed in the main control module.

[0109] It should also be noted that Figure 2 only uses the cache of 128 frames of data as an example. In other embodiments, 56 frames of data or 32 frames can also be cached. The embodiment of the present application does not limit this and can be determined based on demand or hardware support capabilities. In this way, the cached multi-frame data will continue to perform Doppler dimension DC removal and inter-frame FFT to obtain a range-Doppler (RD) diagram.

[0110] In the process shown in Figure 2, some steps can be skipped, such as distance-dimensional DC removal or distance-dimensional DC removal or windowing; some steps can be simplified, such as non-coherent accumulation can be skipped and CFAR detection is performed only on one of the channels, or only some channels are selected for non-coherent accumulation and CFAR detection is performed only on these channels.

[0111] To facilitate those skilled in the art to better understand the process shown in FIG. 2 , the process of different target detection methods will be described below.

[0112] In some embodiments, the process of the target detection method is shown in FIG3 , and includes the following steps:

[0113] Step 201: Perform distance-dimensional FFT on the frame data.

[0114] Step 202: extract target data from the frame data obtained after range-dimensional FFT processing.

[0115] Step 203: windowing the target data according to the sliding window, and performing digital signal processing on the windowed data.

[0116] The length of the sliding window corresponds to a duration of the same order of magnitude as the period of the periodic motion of the target.

[0117] To facilitate those skilled in the art to better understand the target detection method described in the embodiment shown in FIG3 , the steps are explained below.

[0118] In step 201, frame data refers to the data corresponding to the echo signal generated by a frame of detection signal after being reflected by a target. It is understood that target detection by a radar typically has a certain period, as shown in Figure 4. Within a detection period, the radar first transmits a frame of detection signal. Taking frequency modulated continuous wave (FMCW) radar detection as an example, as shown in Figure 5, a frame of detection signal will be composed of multiple chirp signals (the absence of idle periods between chirp signals shown in Figure 5 is only an example; in some cases, there may be idle periods between chirp signals). At the same time as the detection signal begins to be transmitted, the radar prepares to begin receiving the echo signal reflected by the target. The echo signal corresponding to a frame of detection signal undergoes the aforementioned analog-to-digital conversion and range-dimensional DC removal to obtain the aforementioned frame data. Furthermore, each frame of data undergoes a range-dimensional FFT as described above to implement step 201, and this will not be further described here.

[0119] In step 202, the target data is not limited in the embodiment of the present application. It can be any data content belonging to the corresponding frame data. For example, when there are sufficient storage resources and computing power, the target data can be extracted by extracting all the data; or when the real-time performance of target detection is desired to be the highest, at least part of the data can be extracted or data pre-processed on at least part of the data (such as data obtained after pre-processing operations such as complex averaging, complex weighted averaging, or complex summing, etc.). The target data can be extracted according to requirements and hardware conditions. For ease of understanding, the following describes different implementation methods of extracting part of the data as the target data.

[0120] In some embodiments, as shown in FIG6 , target data is extracted from the frame data obtained after the range-dimensional FFT processing by the following steps:

[0121] Step 2021, extract the data corresponding to any one or specified index chirp period (such as the second chirp of each frame) from each frame data obtained after distance-dimensional FFT processing as the target data; wherein, when it is confirmed that the data corresponding to the predetermined rule chirp period is used as the target data, the rules of the chirps selected in each frame are the same.

[0122] In other words, using the data corresponding to a chirp cycle as the representation of frame data makes the processing of target data simple and easy to implement, and the selected data is more realistic, which enables more efficient target detection and is conducive to improving the real-time performance of target detection.

[0123] In some embodiments, as shown in FIG7 , extracting target data from frame data obtained after range-dimensional FFT processing can also be achieved by the following steps:

[0124] Step 2022: Determine the parameters corresponding to each distance unit in each frame of data based on each frame of data obtained after the distance-dimensional FFT processing to generate target data.

[0125] That is to say, multiple data of each distance unit in the frame data are compressed into corresponding parameters so that the overall characteristics of the frame data can be retained, so that the information that can be referenced when performing target detection based on the target data later is more comprehensive and the results obtained are more accurate.

[0126] Of course, the above embodiments are merely provided to reduce the resources occupied by target data. In some embodiments, the target data can also be extracted by considering the validity of the data. For example, radar signals can generally cover a large range, but not all areas covered by the radar are areas of interest. Processing the data corresponding to these areas would result in a waste of resources. Therefore, in some embodiments, as shown in FIG8 , extracting target data from the frame data obtained after range-dimensional FFT processing can also be achieved by the following steps:

[0127] Step 2023: extract data within a preset distance range from the frame data obtained after the distance-dimensional FFT processing as target data.

[0128] That is to say, in the frame data, the range of interest (represented by a preset distance range) is used as the target data. In this way, accurate target detection information can be retained, and the storage resources and computing resources occupied by the target data can be reduced, thereby balancing the accuracy and real-time performance of target detection.

[0129] It should be noted that, in addition to the method shown in FIG8 , in some embodiments, the retention of data within the range of interest can also be achieved by windowing the target data according to a sliding window and performing digital signal processing on the windowed data, as shown in FIG9 , by the following steps:

[0130] Step 2031 : Windowing the target data according to the sliding window, and performing digital signal processing on the windowed data within a preset Doppler range.

[0131] It can also achieve the effect that can be achieved by the embodiment shown in Figure 8, but the dimensions of the two data are different, one is the distance dimension and the other is the Doppler dimension, so the selection method of the range of interest is different. Of course, it is also possible to extract target data in the distance dimension as shown in the embodiment of Figure 7, and to extract data in the Doppler dimension for digital signal processing as shown in the embodiment of Figure 9, etc., which will not be repeated here.

[0132] Of course, the above is only an example of how to implement step 202. In some embodiments, it can also be implemented in other ways. For example, the distance unit with a larger modulus between the distance units (similar to max-pooling) can be used as the target data, etc., which will not be described in detail here.

[0133] In step 203, the windowing method is not limited. For example, the window can be a sliding window, or a sliding window can be a fixed-length window, or a variable-length window, etc., which will not be listed one by one here. For ease of understanding, a fixed-length sliding window can be used as an example for explanation. However, it does not mean that only a fixed-length sliding window can be used. For example, a variable-length window can be changed according to the periodic movement of the target, such as changes in breathing frequency.

[0134] In some embodiments, as shown in FIG10 , for the 1st to 129th frame data generated after the radar is started, taking the window length of the sliding window as 128 frames as an example, the data obtained by the sliding window for the first time is the 1st to 128th frame data (the data in the shaded part of the first row in FIG10 ), and the data obtained for the second time is the 2nd to 129th frame data (the data in the shaded part of the second row in FIG9 ), that is, the sliding step size of the sliding window is 1 frame.

[0135] Of course, the above are only examples. In other embodiments, the window length and sliding step size of the sliding window may also adopt other parameters, which are not listed here one by one.

[0136] Furthermore, in step 203, the data signal processing is not limited. It is understandable that different data may need to be processed for different needs. For example, in some embodiments, target detection may be achieved by simply detecting the target point. In some embodiments, target classification may be required to obtain a more accurate, reliable, and referenceable target, as shown in FIG2 . These are not listed here. The target detection process shown in FIG2 will be used as an example for explanation. However, this does not necessarily mean that target classification must be performed after processing such as CFAR, as shown in FIG2 .

[0137] As shown in FIG2 , the digital signal processing process for target detection generally includes constant false alarm detection, a typical constant false alarm detection method, i.e., constant false alarm detection based on a noise floor. It is understood that, at this point, the more accurate the noise floor estimation, the better the constant false alarm detection effect, and the more accurate the target detection result. Therefore, in some embodiments, as shown in FIG11 , the target data is windowed according to a sliding window, and the windowed data is subjected to digital signal processing, which can be achieved by the following steps:

[0138] Step 2032 : Windowing the target data according to the sliding window, and performing noise floor estimation on the windowed data within a preset range, wherein the upper limit of the preset range is determined according to the noise floor estimation result of the farthest distance unit.

[0139] Step 2034: Perform constant false alarm detection based on the estimated noise floor.

[0140] That is, by setting a preset range to constrain the noise floor estimation, an upper bound is constructed according to the noise floor estimation result of the farthest distance unit, which is conducive to improving the accuracy of noise estimation.

[0141] In some examples, the noise floor is estimated within a preset range using the following expression: i '=min(n i ,α*n g );

[0142] Among them, n i ' is the noise floor estimation result of the i-th distance unit within the preset range, n i is the initial noise floor estimation result of the i-th distance unit, α is the preset parameter, α≥1, n g is the initial noise floor estimation result of the farthest distance unit.

[0143] It should be noted that the preset parameter α can be set and updated based on demand or engineering data, experience, etc.

[0144] It should also be noted that the above is only an exemplary description of noise floor estimation. In some embodiments, the initial noise floor estimation result of the i-th range unit can be directly selected as the noise floor result used in CFAR. Alternatively, the minimum value, maximum value, quantile, average, median, etc. of the initial noise floor estimation results of the last few regions of interest can be selected as the noise floor estimate. Alternatively, when estimating the noise floor, only some Doppler units of a certain range unit can be selected instead of all Doppler units. The noise floor estimation methods are not listed one by one here.

[0145] In addition, the radar shown in Figure 2 is a SISO-combine system. When the radar includes multiple channels, inter-channel integration, such as non-coherent integration or correlation integration, can be performed on the RD spectra of the multiple channels. The NVE module can also be used to estimate the noise floor of each range bin in the accumulated data. The noise floor estimation method can adopt any of the aforementioned noise floor estimation schemes, which will not be detailed here.

[0146] In step 203, the data is windowed and has a certain order. If the order is disrupted, the temporal correlation of the data is destroyed and it can no longer reflect the movement accumulated for a certain period of time, which is not conducive to subsequent target detection. As shown in Figure 2, the data is accumulated in the main control module. When a certain amount of data is accumulated, the baseband module reads the data and continues to process it. In this process, the data writing order and reading order may be different due to the storage method. Therefore, in some embodiments, the data needs to be rearranged to ensure its correct order. For example, in some embodiments, it is assumed that the data is cached using Ring FIFO. At this time, as shown in Figure 12, the target detection method also includes the following steps:

[0147] Step 204: Read data from the Ring FIFO and reorder the read data.

[0148] Thus, the efficiency of subsequent processing can be improved by rearranging data.

[0149] In order to facilitate those skilled in the art to better understand the above data storage and rearrangement, the following will be explained in conjunction with Figure 13.

[0150] As shown in Figure 13, assuming that the data obtained after the baseband module (shown as BB in the figure) performs distance-dimensional FFT on the frame data belongs to the target data in the range of interest (such as from the 5th distance unit to the 20th distance unit), then the main control module (shown as MCU in the figure) will need to accumulate a two-dimensional matrix as shown in Figure 13, where the rows of the matrix correspond to the frame numbers and the columns of the matrix correspond to the distance units.

[0151] Assuming that Ring FIFO is used in the main control module to store the above two-dimensional matrix, the storage process can be as follows:

[0152] Initialize a CPU SRAM with N f Columns, each column can store N r ×N rx ×N tx Number, where N f To design the maximum number of frames for inter-frame FFT, N r 、N rx and N tx The number of range units, transmit (Tx) channels, and receive (Rx) channels of interest are respectively. The buffer can be initialized to all zeros during initialization, and the FIFO head pointer p_head points to the beginning of column 0 of the Ring-FIFO.

[0153] In the 0th frame, the corresponding data (of each distance unit, each Tx channel and each Rx channel) is cached in the 0th column of the Ring-FIFO, and then p_head points to the 1st column.

[0154] Similarly, for the k-th frame data (k=1, 2, . . . , 126), the corresponding data is cached in the k-th column of the Ring-FIFO, and then p_head points to the k+1-th column.

[0155] At the 127th frame, the corresponding data is cached in the 127th column of the Ring-FIFO. Since 128 frames of data are already stored at this point, the data in the Ring-FIFO needs to be moved to the baseband module to perform the aforementioned inter-frame FFT processing. At this time, the data from the column pointed to by p_head is first moved to the 127th column, and then the data from column 0 to the column pointed to by p_head is moved. Obviously, the order of reading data at this time is different from the order of writing, and it needs to be restored. In other words, the data needs to be rearranged so that it is distributed continuously along the frame dimension, thereby improving the computational efficiency of the baseband module.

[0156] After the transfer is complete, the Ring-FIFO pointer p_head returns to column 0. Similarly, during the new data storage and transfer process, the data buffer of frame 128 is stored in column 0 of the Ring-FIFO. During the transfer, the data of frames 1 to 127 are first transferred to the baseband module, and then the data in column 0 is transferred to the baseband module. The same process is repeated for subsequent frames, and will not be repeated here.

[0157] It should be noted that in the above example, inter-frame processing is performed after all valid data is stored in the Ring-FIFO. In other embodiments, after storing partial frame data, such as the 0th to 63rd frame data, the data in the Ring-FIFO can be moved to the baseband module in a similar manner to perform inter-frame FFT, etc., wherein, since the 64th to 127th columns are all initial values ​​assigned during initialization, it can be equivalent to the inter-frame FFT of the baseband module filling the data sequence with zeros. Of course, the above is only an example and does not mean that 128 frame data must be used for inter-frame FFT, or that Ring-FIFO storage must be used, etc., which will not be elaborated here.

[0158] In addition, still as shown in FIG2 , if the user wants to obtain a more intuitive and accurate result, in some embodiments, as shown in FIG14 , after obtaining the target, the following steps are further performed:

[0159] Step 205 : Verify the targets in each preset area according to the targets falling into the preset areas. The preset areas are obtained by dividing the detection space.

[0160] Step 206: Output the detection results of each preset area according to the verified targets.

[0161] That is to say, taking into account that for a target with a certain volume and occupying a certain space, multiple target points will usually be detected, and the currently detected target will be used to continue to be combined with the area division for target verification to improve the detection results. Based on the division of the detection space in the application scenario, the detection results of each preset area are output, so that the target will be presented in the preset area. The distribution is more intuitive and accurate, which is more conducive to users making decisions based on the output results, and the user experience is better.

[0162] To facilitate those skilled in the art to better understand the embodiment shown in FIG. 14 , the steps thereof will be explained below.

[0163] In step 205, the detection space and the preset area are not limited, and they may vary according to different application scenarios, application requirements, etc. For example, in the application scenario of the vehicle-mounted radar, the detection space may be inside the vehicle, and the preset area at this time includes at least any one of the following items: the area corresponding to the seat, the area corresponding to the aisle (foot mat), so that the driver or adults and other subjects can better perceive the situation inside the vehicle. For example, in a factory work scenario, the detection space may be a factory building, and the preset area may include the workstations of each employee to avoid production risks caused by employees not being able to monitor the working conditions of the machines due to employees no longer being at their workstations. They will not be described one by one here. In the following, for ease of understanding, the above-mentioned in-vehicle space will be used as an example for explanation, but this does not mean that the corresponding solution can only be implemented in the vehicle.

[0164] Taking the interior of a vehicle as an example, the interior space is abstracted into a coordinate system as shown in Figure 15. The horizontal axis of this coordinate system represents the azimuth angle relative to the radar installed in the vehicle, and the horizontal axis represents the elevation angle relative to the radar installed in the vehicle. In this case, the vehicle interior is divided into six preset areas, namely the three rear seats and the three aisles in front of the three seats, which are the different fill areas shown in Figure 15. Depending on the requirements, as shown in Figure 15, the different preset areas can overlap or not overlap; some areas in the vehicle can belong to multiple preset areas at the same time, or not belong to any area.

[0165] It should be noted that Figure 15 is only an abstract way of detecting space. In some embodiments, the detection area can be abstracted from all or one of the dimensions of distance, azimuth, and pitch, or all or one of the dimensions of the xyz dimension of the rectangular coordinate system, and further divided into preset areas. They will not be described one by one here.

[0166] In step 205 , the verification method is not limited. For example, the verification may be performed based on the number or proportion of target points.

[0167] For example, in some embodiments, as shown in FIG16 , based on the objects falling within the preset areas, verifying the objects in the preset areas can be achieved through the following steps:

[0168] Step 2051 : Verify the targets in each preset area according to the proportion of the targets falling in each preset area among the detected targets.

[0169] For another example, in some embodiments, as shown in FIG17 , based on the objects falling within the preset areas, verifying the objects in the preset areas can be achieved through the following steps:

[0170] Step 2052: Verify the targets in each preset area according to the signal-to-noise ratio of the targets falling within each preset area.

[0171] Of course, the above are only examples. In some cases, targets at other locations can also be verified based on the target point falling into locations where targets are known to exist, such as the driver's seat. I will not go into details here.

[0172] In step 206, the output method is not limited. The detection results of each preset area can be directly output as a point cloud, or the divided preset areas and the point clouds of each preset area can be simultaneously presented and output, etc., which will not be listed here one by one.

[0173] In order to facilitate those skilled in the art to better understand the above embodiment, the following will be described using proportion verification as an example.

[0174] Assume that a total of 11 target points are detected during a certain target detection process, and the division of the preset area is shown in Figure 15. At the same time, the distribution of the target points in the preset area is shown in Figure 18, that is, the area seat A (6 target points), seat B (2 target points), seat C (0 target points), aisle A (0 target points), aisle B (1 target point) and aisle C (0 target points), where the target points in Figure 18 are represented by solid circles.

[0175] At this point, the algorithm obtained according to the above embodiment will be as follows:

[0176] Initialize the flags flag_region_i of the six regions to be 0, i.e., no one exists (i=0, 1, ..., 5);

[0177] If the total valid target number total_valid_tgt_num is 0, it is determined that there is no one in the car and the process ends. Otherwise, the following operations are performed:

[0178] Traverse all regions. For region i, if the statistical value of the number of valid targets belonging to the region tgt_num_region_i exceeds 25% of the total number of valid targets total_valid_tgt_num, then set the flag flag_region_i of region i to 1;

[0179] If there is a region, if its flag_region_i is 1, then set flag_region_empty to 0, that is, there is someone in the car;

[0180] Traverse the seating areas. If flag_region_i is 1 in any seating area, determine that no one exists in the corresponding aisle area (for example, if it is determined that there is someone in seat A, then determine that no one exists in aisle A).

[0181] Output the results.

[0182] At this time, the output result will be as shown in Figure 18. Area seat A shows 6 target points, and the rest of the preset areas are empty.

[0183] It should be noted that the threshold of 25% in the above example is only an example. In other embodiments, other thresholds or other judgment conditions may be used, which will not be described in detail here.

[0184] It is also understandable that during the detection process, it may be necessary to combine different periodic motions for observation, so that the target detection result can be obtained more accurately. Based on this, in some embodiments, as shown in Figure 19, the target detection method includes the following steps:

[0185] Step 1801: Perform distance-dimensional FFT on the frame data.

[0186] Step 1802: extract target data from the frame data obtained after range-dimensional FFT processing.

[0187] Step 1803 , windowing the target data according to the sliding windows corresponding to the cycles of different vital sign parameters of the living body, and performing digital signal processing on the windowed data corresponding to sliding windows of different lengths.

[0188] Among them, the periods of the vital sign parameters corresponding to different sliding windows are different, and the length of each sliding window has the same order of magnitude as the period of the vital sign parameter.

[0189] Among them, step 1801 and step 1802 are roughly the same as the aforementioned step 201 and step 202. The only difference is that the subsequent steps 1803 and step 203 are based on multiple different periodic motions for detection, while the other describes only one. It can be understood that when detection is based on multiple different periodic motions, after the accumulation of data with a shorter period is completed, the data continues to accumulate until the accumulation of data with a longer period is completed, and the processing of the data is completed. For example, assuming that one periodic motion corresponds to 32 frames and one periodic motion corresponds to 128 frames, both of which need to be observed, then in the target detection process, a detection result is output starting from the completion of the acquisition of the target data corresponding to the 32nd frame data, and a new detection result is continued to be output after the target data corresponding to the 33rd frame data is acquired...until the target data corresponding to the 128th frame data is acquired, two detection results will be output (one is the detection result corresponding to the 32 frame data, and the other is the detection result corresponding to the 128 frame data), or the result after the comprehensive processing of these two results.

[0190] In this way, by obtaining the period of the vital sign parameters of the living body, the length of the sliding window can be adjusted in time, so as to better detect the living target and better monitor the status of the living body (such as children, pets, etc.) in order to monitor and timely deal with the performance of the living body.

[0191] It should be noted that the embodiments of the present application do not limit the vital sign parameters. For example, the vital sign parameters may include respiration. Another example is that the vital sign parameters may include heartbeat and / or pulse. Of course, the above are merely examples. Under other requirements, other vital sign parameters that can reflect motion characteristics may also be used, and they are not listed here one by one.

[0192] Thus, in the above embodiment, the frame-level FFT is adopted to replace the traditional chirp-level Doppler FFT to achieve the effective accumulation of the target's motion energy and achieve the purpose of accurate detection and measurement, and the sliding window FFT processing can be performed on multiple frames to further improve the result update frequency; by performing post-processing methods such as regional statistics on the target points detected by the radar, it is possible to determine whether there are people in each area.

[0193] In the above embodiment, the target's speed is determined by using a frame-level FFT instead of the traditional chirp-level Doppler dimension FFT to achieve accurate detection of the target of interest, and the sliding window FFT processing can be performed on multiple frames to further increase the update frequency of the results, thereby improving the real-time performance of the system. At the same time, the amount of data processed can be reduced by processing only the distance and / or Doppler area of ​​interest. In addition, when estimating the background noise, the background noise is obtained by taking the minimum value based on the estimate of the current distance unit (bin) combined with the estimate of the last distance unit of interest, which can be more suitable for application scenarios in relatively sealed environments such as the interior of a car. When judging the regional logic, by counting the target points detected in each frame according to the preset area design and judging the counting results according to the preset rules, it is possible to accurately judge whether there is a target in each preset area.

[0194] Figure 20 is a flow chart of target detection based on per-channel in an embodiment of the present application. As shown in Figure 20, based on the process structure shown in Figure 1, the RD spectrum of multiple channels is incoherently accumulated into single-channel data for CFAR. In this embodiment, all channels are CFAR processed one by one, and then the binary integration in channel domain is processed, and then the horizontal / azimuth and pitch angles are estimated. For example, for a system with M channels, after multiple channels are subjected to inter-frame FFT as shown in Figure 1, CFAR processing is performed on multiple channels to obtain M bit masks, each element of which can be of bool type, and each range unit (range bin) and / or Doppler unit (Doppler bin) is judged to determine whether there is a target (if it is defined as 1, it is defined as present, and if it is 0, it is defined as not present). At the same time, these M bit masks can be accumulated, and the range of each element can be defined as 0 to (M-1). The accumulated mask can be subsequently tested for a second time, that is, each element of the accumulated mask is compared with a preset threshold M0. If the value x of the unit to be tested (such as an element of the accumulated mask) is i >M0, then it can be output that there is a target in the unit to be detected. If x i≤M0, it can be output that there is no target in the unit to be detected. i , M, M0 are positive integers.

[0195] In the detection method process shown in FIG20 , since each channel is subjected to CFAR processing separately, the performance loss caused by channel imbalance can be effectively reduced, and the method has better robustness in practical applications.

[0196] The present disclosure also provides another target detection method, as shown in FIG21 , including the following steps:

[0197] Step 101 : Based on the range-Doppler spectrum, CFAR processing is performed on at least two transceiver channels separately to obtain candidate target data of the at least two transceiver channels.

[0198] Step 102 : Process candidate target data based on at least two transceiver channels to obtain final target data.

[0199] In this way, the at least two transceiver channels are subjected to separate constant false alarm detection, so that the results of the at least two transceiver channels do not interfere with each other and do not affect other channels. Therefore, even if an abnormal channel or abnormal data exists between the at least two transceiver channels, it will not affect the processing of other normal channels. Subsequently, processing is performed based on the candidate target data of the at least two transceiver channels to obtain the final target data, which supports the detection of the two transceiver channels. It not only ensures the normal processing of the abnormal channel or the channel to which the abnormal data belongs, but also avoids the interference of the abnormal channel or the channel to which the abnormal data belongs on other channels, alleviates the problem of imbalance between channels, and the problem of different delays and phases between channels due to inconsistent lengths of the transceiver to the antenna and inaccurate compensation, thereby improving the accuracy of detection.

[0200] To help understand the target detection method provided by the above embodiment, its steps will be explained and illustrated below.

[0201] In step 101 , the number of transceiver channels that are individually subjected to CFAR processing is not limited, and may be, for example, 2, 3, 5, or all transceiver channels.

[0202] For example, in some embodiments, all transceiver channels can be individually subjected to CFAR processing. This completely avoids interference between the channels and improves accuracy. At this point, as shown in Figure 22, at least two transceiver channels are subjected to CFAR processing based on the range-Doppler spectrum. This is achieved by: Step 201, based on the range-Doppler spectrum, each transceiver channel is subjected to CFAR processing. Correspondingly, candidate target data based on the at least two transceiver channels is processed. This is achieved by: Step 202, the candidate target data of each transceiver channel is processed.

[0203] Of course, the above is only an example. In some embodiments, CFAR detection may be performed on some transceiver channels separately to avoid the adverse effects of channels with relatively poor signal-to-noise ratio or severe transceiver leakage (TRX leakage). Details will not be given here.

[0204] In step 101, the constant false alarm processing method of a single transceiver channel is not limited. For example, in some embodiments, it can be implemented based on existing constant false alarm processing, except that the original data accumulation of multiple transceiver channels is changed to data accumulation of a single transceiver channel.

[0205] In some embodiments, as shown in FIG23 , performing constant false alarm processing on at least two transceiver channels based on the range-Doppler spectrum can be achieved by the following steps:

[0206] Step 1011 , based on the range-Doppler spectrum, the noise floor of the transceiver channel that is subjected to the CFAR processing is estimated separately to obtain a noise floor estimation result of the single transceiver channel.

[0207] Step 1012 : Based on the noise floor estimation result of each transceiver channel, a constant false alarm detection is performed on the corresponding transceiver channel to obtain candidate target data corresponding to each transceiver channel.

[0208] That is, when performing constant false alarm processing on a single transceiver channel, the noise floor estimation is performed on the single transceiver channel as an independent whole, so as to implement CFAR based on the noise floor estimation based on the noise floors estimated by each channel.

[0209] In step 1011, the noise floor estimation method is also not limited. For example, in some embodiments, based on the range-Doppler spectrum, the noise floor estimation for a transceiver channel that is subjected to CFAR processing can be performed separately. This can be achieved by performing noise floor estimation for the transceiver channel that is subjected to CFAR processing within a preset range based on the range-Doppler spectrum to obtain a noise floor estimation result for the individual transceiver channel. The upper bound of the preset range is determined based on the noise floor estimation result of the farthest range unit. In other words, by constraining the noise floor estimation within the preset range, the upper bound is constructed based on the noise floor estimation result of the farthest range unit, which helps improve the accuracy of the noise estimation.

[0210] In some examples, the noise floor is estimated for each transmit and receive channel separately, using the following expression: i '=min(n i ,α*n g );

[0211] Among them, n i' is the noise floor estimation result of the i-th distance unit of a single transceiver channel within the preset range, n i is the initial noise floor estimation result of the i-th range unit on a single transceiver channel determined according to the range-Doppler spectrum, α is a preset parameter, α≥1, n g It is the initial noise floor estimation result of the farthest distance unit on a single transmit and receive channel.

[0212] The preset parameter α can be set and updated based on demand, engineering data, experience, etc.

[0213] It should also be noted that the above is merely an exemplary description of noise floor estimation. In some embodiments, the initial noise floor estimation result of the i-th range unit on a single transceiver channel may be directly selected as the noise floor result used when performing CFAR processing on the transceiver channel. Alternatively, the minimum value, maximum value, quantile, average, median, etc. of the initial noise floor estimation results of the last few regions of interest on a single transceiver channel may be selected as the noise floor estimate. Alternatively, when estimating the noise floor, only some Doppler units of a certain range unit may be selected instead of all Doppler units. All noise floor estimation methods will not be listed here.

[0214] In step 1012, constant false alarm detection can be implemented by threshold detection. For example, the noise floor estimation result of a single transceiver channel is composed of the noise floor estimation results of each range unit on the single transceiver channel, and the constant false alarm detection result corresponding to the single transceiver channel is composed of the constant false alarm detection results of each Doppler unit of each range unit on the single transceiver channel.

[0215] Based on the noise floor estimation result of a single transceiver channel, constant false alarm detection is performed on the corresponding transceiver channel separately, which is achieved through the following expression:

[0216] Where F(c,r,v) is the constant false alarm detection result of the vth Doppler unit of the rth range unit of the cth channel, P(c,r,v) is the echo energy at the vth Doppler unit of the rth range unit of the cth channel, n′(c,r) is the noise floor estimation result of the rth range unit of the cth channel, and β is a preset parameter, β ≥ 1.

[0217] From the above expression, we can see that when F(c,r,v) is 1, it indicates that a target is present in the vth Doppler bin of the rth range bin, and when F(c,r,v) is 0, it indicates that a target is not present in the vth Doppler bin of the rth range bin. In other words, binarizing the detection results of a single transceiver channel eliminates interference from specific values ​​(especially extreme data) during the detection process, improving detection accuracy.

[0218] The above expression only provides an explanation by taking binarization of the result to obtain candidate target data as an example. In some embodiments, quantization methods other than binarization can be used to quantize the result, which can also reduce the interference of extreme data to a certain extent, and it is not necessary to use binarization to achieve it. They will not be listed one by one here.

[0219] It should also be noted that the above is only an example of constant false alarm processing for a single transceiver channel provided in combination with CFAR based on noise floor estimation, but it does not mean that the constant false alarm processing for a single transceiver channel in the embodiment of the present application can only be implemented as above. For example, it can be implemented in combination with CA (Cell Averaging)-CFAR, OS-CFAR, GO-CFAR, etc. During implementation, the processing of multiple transceiver channels in the original algorithm can be regarded as a single transceiver channel, and no further details will be given here.

[0220] In step 102, processing candidate target data from at least two transceiver channels can be considered as further performing target verification based on candidate target data from the transceiver channels, building upon the target verification performed in step 101 based on data from a single transceiver channel. This can improve target accuracy and reliability. Therefore, step 102 can be implemented using the concept of constant false alarm detection.

[0221] For example, in some embodiments, as shown in FIG23 , processing the candidate target data of each transceiver channel to obtain the final target data can be achieved by the following steps:

[0222] Step 1021 , performing constant false alarm detection based on the currently acquired candidate target data and a preset threshold to obtain final target data.

[0223] In other words, based on the existing constant false alarm processing, which uses non-correlated accumulated results to perform threshold judgment, the candidate target data is processed with a similar threshold to obtain the final detection result. This is difficult to implement, highly efficient, and can improve detection accuracy, real-time performance, and user experience.

[0224] At this time, constant false alarm detection is performed based on the currently acquired candidate target data and the preset threshold, which is achieved through the following expression:

[0225] Where F(c,r,v) is the constant false alarm detection result of the vth Doppler unit of the rth range unit of the cth channel, N c is the total number of channels, F0 is the preset threshold, and F′(r,v) is the constant false alarm detection result of the vth Doppler unit in the rth range unit.

[0226] That is to say, the candidate target data of the transmitting and receiving channels are accumulated, and then it is judged whether the accumulated result exceeds the preset threshold to determine the final target data. The final target data is used to characterize whether there is a target on the vth Doppler unit of the rth range unit, thereby avoiding false monitoring and missed detection caused by multipath, etc., and improving the accuracy and reliability of target detection.

[0227] Of course, the above is merely an exemplary explanation of obtaining the final target data by means of a threshold value. In some embodiments, if the constant false alarm detection result of the vth Doppler unit of the rth range unit of the cth channel is represented by a quantized numerical value, then obtaining the final target data can be determined by determining on how many channels the constant false alarm detection result of the vth Doppler unit of the rth range unit exceeds a preset value; alternatively, the above accumulation process can be replaced by a weighted summation or averaging process, etc., which will not be described in detail here.

[0228] It should be noted that the above embodiment only limits the target verification in the target detection process (i.e., obtaining results similar to the existing CFAR processing), but does not limit other data processing processes in the target detection process. It can be understood that the embodiments shown in Figures 21-23 can be further combined with other data processing implementation schemes, for example, combined with a multi-frame joint processing scheme, etc.

[0229] To facilitate understanding of the above-mentioned target detection scheme and multi-frame joint processing scheme, the multi-frame joint processing scheme is first explained below.

[0230] As shown in FIG2 , a radar (taking a frequency modulated continuous wave (FMCW) radar as an example) receives an echo signal through an antenna and performs analog-to-digital conversion, sampling, and other processing to obtain a digital signal. The digital signal is then processed by a main control module (e.g., a microcontroller unit (MCU)) and a baseband (BB) module (e.g., a baseband chip) as shown in FIG2 to achieve target detection. For example, in FIG2 , the digital signal will be sequentially subjected to range DC removal and range Fourier transform (FFT, also expressed as 1D-FFT or fast time FFT) in the baseband module. Subsequently, the signal processed by the baseband module will be sent to the main control module for multi-frame accumulation. When the data has accumulated to a preset number of frames, such as 128 frames, the main control module will transfer the accumulated multi-frame data from the CPU static random-access memory (SRAM) to the baseband module. Doppler DC removal, frame FFT, and inter-channel accumulation (e.g., non-coherent integration, correlation accumulation, etc.) will then be further processed in the baseband module. Constant false alarm detection (e.g., peak-based CFAR detection) will then be performed on the noise obtained by noise estimation (e.g., using a noise variance estimator) in the baseband module. Finally, angle detection and target classification are performed to obtain target distance, speed, and / or angle information. When performing angle detection, azimuth and elevation estimation can be performed based on the CFAR results. This can be achieved, for example, through technologies such as digital beam forming (DBF) and direction of arrival estimation.

[0231] The implementation of the digital signal processing shown in FIG2 is described above.

[0232] Therefore, it can be seen from the above that, in combination with the above-described multi-frame joint processing scheme, that is, using multi-frame joint technology when acquiring the RD spectrum, or in other words, replacing the non-correlated accumulation and CFAR processing in the above-described multi-frame joint technology scheme with the target detection method provided by the embodiments shown in Figures 21-23, that is, combining the scheme provided by the above-described embodiment with the process shown in Figure 2, the target detection process will be as shown in Figure 24. In this case, after CFAR processing is performed on all channels one by one, binary integration in the channel domain is continued, and then the horizontal / azimuth and pitch angles are estimated. For example, for a system with M channels, after performing the inter-frame FFT as shown in Figure 2, CFAR processing is performed on each channel to obtain M bit masks, each element of which can be of bool type, to determine whether each range bin and / or Doppler bin determines whether a target exists (if it is 1, it is defined as present, and if it is 0, it is defined as not present). At the same time, these M bit masks can be accumulated, where the range of each element can be defined as 0 to (M-1). Subsequently, the accumulated mask can be subjected to secondary detection, that is, each element of the accumulated mask can be compared with a preset threshold value M0. If the value xi of the unit to be detected (such as an element of the accumulated mask) is greater than M0, then it can be output that there is a target in the unit to be detected (that is, the final target data indicates that there is a target); if xi≤M0, then it can be output that there is no target in the unit to be detected (that is, the final target data indicates that there is no target). Among them, xi, M, and M0 are positive integers. Among them, in the detection method process shown in Figure 24, since each channel is used to perform CFAR processing separately, it can effectively reduce the performance loss of channel imbalance and has better robustness in practical applications.

[0233] To facilitate those skilled in the art to better understand the process shown in FIG. 24 , the process of different target detection methods will be described below.

[0234] In some embodiments, the process of the target detection method is shown in FIG25 , and further includes the following steps:

[0235] Step 103: Perform distance-dimensional FFT on the frame data.

[0236] Step 104: extract target data from the frame data obtained after the range-dimensional FFT processing.

[0237] Step 105 , performing sliding windowing on the target data, and performing 2D FFT on the windowed data to obtain a range-Doppler spectrum.

[0238] The length of the sliding window corresponds to a duration of the same order of magnitude as the period of the periodic motion of the target.

[0239] To help those skilled in the art better understand the target detection method described in the embodiment shown in Figure 25, its steps will be explained below.

[0240] In step 103, frame data refers to the data corresponding to the echo signal generated after a frame of detection signal is reflected by the target. It is understood that target detection by a radar typically has a certain period, as shown in Figure 4. Within a detection period, the radar first transmits a frame of detection signal. Taking frequency modulated continuous wave (FMCW) radar detection as an example, as shown in Figure 5, a frame of detection signal will be composed of multiple chirp signals. Simultaneously with the start of the detection signal transmission, the radar prepares to receive the echo signal reflected by the target. The echo signal corresponding to a frame of detection signal undergoes the aforementioned analog-to-digital conversion and range-dimensional DC removal to obtain the aforementioned frame data. Furthermore, each frame of data undergoes a range-dimensional FFT as described above to implement step 103, and this will not be further described here.

[0241] In step 104, the target data is not limited in the embodiments of the present application and can be any data content belonging to the corresponding frame data. For example, when storage resources and computing power are sufficient, the target data can be extracted by extracting all the data; or when the real-time performance of target detection is desired, partial data can be extracted. The target data can be extracted based on requirements and hardware conditions. For ease of understanding, the following describes different implementations of extracting partial data as target data.

[0242] In some embodiments, target data is extracted from frame data obtained after distance-dimensional FFT processing by extracting data corresponding to a chirp cycle from each frame data obtained after distance-dimensional FFT processing as target data.

[0243] In other words, using the data corresponding to a chirp cycle as the representation of frame data makes the processing of target data simple and easy to implement, and the selected data is more realistic, which enables more efficient target detection and is conducive to improving the real-time performance of target detection.

[0244] In some embodiments, extracting target data from frame data obtained after distance-dimensional FFT processing can also be achieved through the following steps: based on each frame data obtained after distance-dimensional FFT processing, determining the parameters corresponding to each distance unit in each frame data to generate target data.

[0245] That is to say, multiple data of each distance unit in the frame data are compressed into corresponding parameters so that the overall characteristics of the frame data can be retained, so that the information that can be referenced when performing target detection based on the target data later is more comprehensive and the results obtained are more accurate.

[0246] Of course, the above embodiments merely provide a means of reducing the resources occupied by target data. In some embodiments, target data can also be extracted based on data availability. For example, radar signals typically cover a large area, but not all areas covered by the radar are areas of interest. Processing data corresponding to these areas would result in a waste of resources. Therefore, in some embodiments, extracting target data from frame data obtained after range-dimensional FFT processing can also be achieved by extracting data within a preset distance range from the frame data obtained after range-dimensional FFT processing as target data.

[0247] That is to say, in the frame data, the range of interest (represented by a preset distance range) is used as the target data. In this way, accurate target detection information can be retained, and the storage resources and computing resources occupied by the target data can be reduced, thereby balancing the accuracy and real-time performance of target detection.

[0248] It should be noted that, in addition to being achieved through the above-described method, in some embodiments, the retention of data within the range of interest can also be achieved through the steps of windowing the target data according to a sliding window and performing digital signal processing on the data within a preset Doppler range in the windowed data. Windowing the target data according to a sliding window and performing digital signal processing on the windowed data can also achieve the effect of retaining the data in the region of interest, but the dimensions of the two data are different, one is the distance dimension and the other is the Doppler dimension. Therefore, the selection method of the range of interest is different. Of course, it is also possible to extract target data in the distance dimension and extract data in the Doppler dimension for digital signal processing, etc., which will not be elaborated here.

[0249] Of course, the above is only an example of how to implement step 104. In some embodiments, it can also be implemented in other ways. For example, the distance unit with a larger modulus between the distance units (similar to max-pooling) can be used as the target data, etc., which will not be described in detail here.

[0250] In step 105, the sliding windowing method is not limited. For example, the sliding window can be a fixed-length window; the sliding window can also be a variable-length window, etc., which will not be listed one by one here. For ease of understanding, the fixed-length sliding window can be used as an example for explanation. However, it does not mean that only a fixed-length sliding window can be used. For example, a variable-length window can be changed according to the periodic motion of the target, such as changes in breathing frequency.

[0251] In some embodiments, as shown in FIG10 , for the 1st to 129th frame data generated after the radar is started, taking the window length of the sliding window as 128 frames as an example, the data obtained by the sliding window for the first time is the 1st to 128th frame data (the data in the shaded part of the first row in FIG9 ), and the data obtained for the second time is the 2nd to 129th frame data (the data in the shaded part of the second row in FIG10 ), that is, the sliding step size of the sliding window is 1 frame.

[0252] Of course, the above are only examples. In other embodiments, the window length and sliding step size of the sliding window may also adopt other parameters, which are not listed here one by one.

[0253] In step 105, when windowing the data, the data needs to have a certain order. If the order is disrupted, the temporal correlation of the data is destroyed, and it can no longer reflect the movement accumulated for a certain period of time, which is not conducive to subsequent target detection. As shown in Figure 24, the data accumulates in the main control module. When a certain amount of data is accumulated, the baseband module reads the data and continues to process it. In this process, the data writing order and reading order may be different due to the storage method. Therefore, in some embodiments, the data needs to be rearranged to ensure its correct order. For example, in some embodiments, it is assumed that the data adopts Ring FIFO cache. At this time, as shown in Figure 26, the target detection method also includes the following steps:

[0254] Step 106: Read data from the Ring FIFO and reorder the read data.

[0255] Thus, the efficiency of subsequent processing can be improved by rearranging data.

[0256] In order to facilitate those skilled in the art to better understand the above data storage and rearrangement, the following will be explained in conjunction with Figure 13.

[0257] As shown in Figure 13, assuming that the data obtained by the baseband module (shown as BB in the figure) after performing a range-dimensional FFT on the frame data is within the range of interest (e.g., from the 5th to the 20th range unit) is the target data, the main control module (shown as MCU in the figure) will need to accumulate a two-dimensional matrix as shown in Figure 13, with the rows of the matrix corresponding to the frame sequence number and the columns of the matrix corresponding to the range unit. The detailed storage process is described above and will not be repeated here.

[0258] In step 105, it is also understood that during the detection process, multi-frame joint detection may require observation in conjunction with different periodic motions, which can more accurately obtain target detection results. Based on this, sliding windowing is performed on the target data, and a 2D FFT is performed on the windowed data. This can be achieved by sliding windowing the target data according to the period of different vital sign parameters of the living subject, and digital signal processing is performed on the windowed data corresponding to sliding windows of different lengths.

[0259] Among them, the periods of the vital sign parameters corresponding to different sliding windows are different, and the length of each sliding window has the same order of magnitude as the period of the vital sign parameters. At this point, it can be understood that when detection is performed based on a variety of different periodic motions, after the accumulation of data with a shorter period is completed, the data continues to accumulate until the accumulation of data with a longer period is completed, and the processing of the data is completed. For example, assuming that one periodic motion corresponds to 32 frames and one periodic motion corresponds to 128 frames, both of which need to be observed, then in the process of target detection, a detection result is output starting from the completion of the acquisition of the target data corresponding to the 32nd frame data, and a new detection result is output after the target data corresponding to the 33rd frame data is acquired... until the acquisition of the target data corresponding to the 128th frame data is completed, two detection results will be output (one is the detection result corresponding to the 32 frame data, and the other is the detection result corresponding to the 128 frame data), or the result after the comprehensive processing of these two results.

[0260] In this way, by obtaining the period of the vital sign parameters of the living body, the length of the sliding window can be adjusted in time, so as to better detect the living target and better monitor the status of the living body (such as children, pets, etc.) in order to monitor and timely deal with the performance of the living body.

[0261] It should be noted that the embodiments of the present application do not limit the vital sign parameters. For example, the vital sign parameters may include respiration. Another example is that the vital sign parameters may include heartbeat and / or pulse. Of course, the above are merely examples. Under other requirements, other vital sign parameters that can reflect motion characteristics may also be used, and they are not listed here one by one.

[0262] In addition, it is also understandable that if the user wishes to obtain more intuitive and accurate results, further processing can be performed after obtaining the target data, such as target classification, further verification, etc.

[0263] Based on this, in some embodiments, as shown in FIG27 , after obtaining the target, the following steps are further performed:

[0264] Step 107 : Verify the targets in each preset area according to the targets falling into the preset areas. The preset areas are obtained by dividing the detection space.

[0265] Step 108: Output the detection results of each preset area according to the verified targets.

[0266] That is to say, taking into account that for a target with a certain volume and occupying a certain space, multiple target points will usually be detected, and the currently detected target will be used to continue to be combined with the area division for target verification to improve the detection results. Based on the division of the detection space in the application scenario, the detection results of each preset area are output, so that the target will be presented in the preset area. The distribution is more intuitive and accurate, which is more conducive to users making decisions based on the output results, and the user experience is better.

[0267] To facilitate those skilled in the art to better understand the embodiment shown in FIG. 27 , the steps thereof will be explained below.

[0268] In step 107, the detection space and the preset area are not limited, and they may vary according to different application scenarios, application requirements, etc. For example, in the application scenario of vehicle-mounted radar, the detection space may be inside the vehicle. In this case, the preset area includes at least any one of the following: the area corresponding to the seat and the area corresponding to the foot mat, so that the driver or an adult or other object can better perceive the situation inside the vehicle. For example, in a factory work scenario, the detection space may be a factory building. In this case, the preset area may include the workstations of each employee to avoid production risks caused by employees not being able to monitor the working conditions of the machine due to employees not being at their workstations. They will not be described one by one here. In the following, for ease of understanding, the above-mentioned in-vehicle space will be used as an example for explanation, but this does not mean that the corresponding solution can only be implemented in the vehicle.

[0269] Taking the interior of a vehicle as an example, the interior space is abstracted into a coordinate system as shown in Figure 15. The horizontal axis of this coordinate system represents the azimuth angle relative to the radar installed in the vehicle, and the horizontal axis represents the elevation angle relative to the radar installed in the vehicle. In this case, the vehicle interior is divided into six preset areas, namely the three rear seats and the three aisles in front of the three seats, which are the different fill areas shown in Figure 15. Depending on the requirements, as shown in Figure 15, the different preset areas can overlap or not overlap; some areas in the vehicle can belong to multiple preset areas at the same time, or not belong to any area.

[0270] It should be noted that Figure 15 is only an abstract way of detecting space. In some embodiments, the detection area can be abstracted from all or one of the dimensions of distance, azimuth, and pitch, or all or one of the dimensions of the xyz dimension of the rectangular coordinate system, and further divided into preset areas. They will not be described one by one here.

[0271] In step 108 , the verification method is not limited. For example, the verification may be performed based on the number or proportion of target points.

[0272] For example, in some embodiments, based on the targets falling into each preset area, verifying the targets in each preset area can be achieved by the following steps: verifying the targets in each preset area based on the proportion of targets falling into each preset area in the detected targets.

[0273] For another example, in some embodiments, verifying the targets in each preset area based on the targets falling within the preset areas can be achieved by the following steps: verifying the targets in each preset area based on the signal-to-noise ratio of the targets falling within the preset areas.

[0274] Of course, the above are only examples. In some cases, targets at other locations can also be verified based on the target point falling into locations where targets are known to exist, such as the driver's seat. I will not go into details here.

[0275] In step 108, the output method is not limited. The detection results of each preset area can be directly output as a point cloud, or the divided preset areas and the point clouds of each preset area can be simultaneously presented and output, etc., which will not be listed here one by one.

[0276] In order to facilitate those skilled in the art to better understand the above embodiment, the following will be described using proportion verification as an example.

[0277] Assume that a total of 11 target points are detected during a certain target detection process, and the division of the preset area is shown in Figure 15. At the same time, the distribution of the target points in the preset area is shown in Figure 18, that is, the area seat A (6 target points), seat B (2 target points), seat C (0 target points), aisle A (0 target points), aisle B (1 target point) and aisle C (0 target points), where the target points in Figure 18 are represented by solid circles.

[0278] At this point, the algorithm obtained according to the above embodiment will be as follows:

[0279] Initialize the flags flag_region_i of the six regions to be 0, i.e., no one exists (i=0, 1, ..., 5);

[0280] If the total valid target number total_valid_tgt_num is 0, it is determined that there is no one in the car and the process ends. Otherwise, the following operations are performed:

[0281] Traverse all regions. For region i, if the statistical value of the number of valid targets belonging to the region tgt_num_region_i exceeds 25% of the total number of valid targets total_valid_tgt_num, then set the flag flag_region_i of region i to 1;

[0282] If there is a region, if its flag_region_i is 1, then set flag_region_empty to 0, that is, there is someone in the car;

[0283] Traverse the seating areas. If flag_region_i is 1 in any seating area, determine that no one exists in the corresponding aisle area (for example, if it is determined that there is someone in seat A, then determine that no one exists in aisle A).

[0284] Output the results.

[0285] At this time, the output result will be as shown in Figure 28. Only the seat area A shows 6 target points, and the rest of the preset areas are empty.

[0286] It should be noted that the threshold of 25% in the above example is only an example. In other embodiments, other thresholds or other judgment conditions may be used, which will not be described in detail here.

[0287] In this way, while performing constant false alarm processing on a single transceiver channel to avoid interference from abnormal channels or data, a frame-level FFT is used instead of the traditional chirp-level Doppler FFT to effectively accumulate the target's motion energy, achieving accurate detection and measurement. Sliding window FFT processing can be performed on multiple frames to further increase the frequency of result updates. Post-processing methods such as regional statistics are performed on target points detected by the radar to determine whether a person is present in each area. For example, a frame-level FFT is used instead of the traditional chirp-level Doppler FFT to determine the target's velocity, achieving accurate detection of targets of interest. Sliding window FFT processing can be performed on multiple frames to further increase the frequency of result updates, thereby improving the system's real-time performance. Furthermore, the amount of data processed can be reduced by processing only the range and / or Doppler region of interest. Furthermore, when estimating background noise, minimizing the background noise based on the estimate of the current range bin combined with the estimate of the last range bin of interest is more suitable for applications in relatively sealed environments such as vehicle cabins. When judging regional logic, by counting the target points detected in each frame according to the preset area design and judging the counting results according to the preset rules, it is possible to accurately judge whether there is a target in each preset area.

[0288] FIG29 is a flow chart of target detection based on frame-level FFT combined with DAE CFAR in an embodiment of the present application. As shown in FIG29 , based on the flow structure shown in FIG1 , based on the frame-dimensional FFT (frame-FFT), DAE CFAR (Constant False Alarm Rate) is used to estimate the azimuth and elevation angles for CFAR data. For example, after performing Doppler dimension CFAR detection on the inter-frame FFT data, azimuth dimension DBF (Azimuth DBF) is performed, and the azimuth CFAR processing (Azimuth CFAR) is continued in combination with the NVE (noise variance estimator, noise variance estimation) noise floor estimation; similarly, the elevation dimension DBF (Elevation DBF) is performed on the azimuth CFAR data, and the elevation CFAR processing (Elevation CFAR) is continued in combination with the NVE noise floor estimation, thereby obtaining the azimuth and elevation information of the target.

[0289] Figure 30 is a schematic diagram of azimuth or elevation CFAR based on the target detection process shown in Figure 29. As shown in Figure 30, the NVE engine can calculate the azimuth / elevation power data using the HIST method, and the azimuth / elevation DBF noise floor estimation can be performed by selecting a rank value (such as the mean or median). When performing DBF, all peaks in the DBF spectrum are obtained, and the SNR of all peak points and the estimated noise floor are used to determine whether to output them as peaks.

[0290] For example, when performing DBF, the amplitudes of all peak points in the DBF spectrum, including the global maximum (Global Max) peak and the local maximum (Local Max) peak, can be obtained. For any peak, if its SNR exceeds a predetermined threshold, the difference between its amplitude (Local Max) and the global maximum amplitude (Global Max) is within a preset range, and the number of peaks already output is less than a preset number, then the peak is output as the true target point. Compared to directly using the peaks in the Doppler spectrum to find the target, since the Doppler spectrum may contain clutter, resulting in the peaks not reflecting the true target, the above method is more accurate in finding the true target point. Secondly, there may be multiple targets in the Doppler spectrum. If only the peaks are used to find the target, the target will be lost. However, the target detection method of this embodiment takes into account the relationship between the candidate target and the global maximum amplitude, thus achieving accurate multi-target detection.

[0291] The present disclosure provides a target detection method, as shown in FIG31 , including:

[0292] Step 10: After performing range-dimensional FFT processing on the echo signal, 1D-FFT data is obtained, and multi-frame joint processing is performed on the 1D-FFT data to obtain an RD spectrum; the RD spectrum is, for example, a two-dimensional range-doppler graph, which contains range Doppler data, that is, two-dimensional range-velocity data.

[0293] Step 20: Detect the target in the region of interest based on the RD map.

[0294] The above detection of the target in the region of interest includes at least one of the following operations: judgment, positioning, identification, etc.

[0295] The target detection method and related devices provided in the embodiments of the present invention can realize target detection solutions in sealed space areas such as cabins, indoors, and factories through multi-frame joint processing technology, that is, through inter-frame accumulation, so as to effectively improve the detection rate, reduce the number of false alarm targets, and greatly improve the accuracy of angle estimation, etc., thereby enabling special targets or weak targets such as infants and young children in the cabin to be accurately detected, realizing applications such as CPD (Child Presence Detection, child detection in the car) and SBR (Safety Belt Reminder, seat belt reminder device).

[0296] An embodiment of the present disclosure also provides a target detection method, which may include: performing a range-dimensional FFT within a chirp and a speed-dimensional FFT between frames on an echo signal to obtain range-speed dimension data; performing a first constant false alarm processing on the range-speed dimension data to obtain a candidate target detection point; performing a second constant false alarm processing on the candidate target detection point, and performing target detection based on the second constant false alarm processing result.

[0297] In the embodiment of the present application, two constant false alarm processes are used to effectively eliminate false alarms and effectively improve detection and measurement performance. This allows special targets or weak targets such as infants in the cabin to be accurately detected, realizing applications such as CPD and SBR.

[0298] In an exemplary embodiment, the step of obtaining the range-velocity dimension data includes: performing range-dimensional FFT processing on the echo signal within the chirp to obtain 1D-FFT data; accumulating frame data on the 1D-FFT data until a predetermined amount of data is accumulated, and then performing inter-frame velocity-dimensional FFT processing to obtain an RD spectrum, i.e., the range-velocity dimension data. For example, a sliding window method can be used to read a predetermined number of frames of 1D-FFT data each time to perform inter-frame FFT processing.

[0299] During the implementation process, accurate detection of targets in the cabin can be achieved based on multi-frame joint processing technology. The multi-frame joint processing solution can be to obtain the range-Doppler map by performing a sliding window FFT on the multi-frame data, or to perform CFAR, DOA and other processing operations on the area of ​​interest after using FIR (Finite Impulse Response) or other complex time-frequency transformation processing. It can also be combined with the set area judgment logic and area parameters to achieve the detection and positioning of living targets such as adults, children, pets, or other non-living targets. CFAR can be NR-CFAR in the Doppler dimension, or RD-CFAR, or DAE (Doppler-Azimuth-Elevation)-CFAR, etc. After CFAR and DoA, post-processing such as clustering, false alarm suppression, and multiple point cloud association can be used. The regional parameters can be determined by adopting at least one of the following schemes, or a combination of at least two schemes, such as clustering operation on the point cloud detected after a certain sliding window multi-frame processing, outlier detection and elimination operation on the point cloud detected after a certain sliding window multi-frame processing, etc.

[0300] In an exemplary embodiment, the step of performing the first CFAR processing includes: performing incoherent accumulation on RD spectra of multiple channels, and performing the first CFAR processing based on noise estimation based on the incoherent accumulation results to obtain candidate target detection points.

[0301] Optionally, the first constant false alarm processing includes: performing incoherent accumulation on the RD spectra of multiple channels, estimating the noise floor of each distance unit after the incoherent accumulation, obtaining a noise floor estimation value of each distance unit, and performing incoherent constant false alarm processing using the noise floor estimation value, wherein, when estimating the noise floor of each distance unit after the incoherent accumulation, the noise floor of each distance unit is adjusted using the global noise floor.

[0302] An optional adjustment method includes adjusting the noise floor of each distance unit in the following manner: i ′=min(n i ,α*n g ), where n i is the original noise floor estimate of the i-th distance unit, and n g is the mean of the noise floor estimates of multiple distance cells, n i ′ is the adjusted noise floor estimate of the i-th range unit, and α> 1. By adjusting the noise floor, we can avoid the target being undetected due to the noise floor being too high.

[0303] In an exemplary embodiment, the second constant false alarm processing includes performing DAE (Doppler-Azimuth-Elevation) CFAR processing on the azimuth angle and the elevation angle respectively.

[0304] In an exemplary embodiment, the step of performing the second constant false alarm processing includes: performing digital beamforming processing on the candidate target detection points, and then performing the second constant false alarm processing based on noise estimation to obtain a screening result in which false target points are filtered out.

[0305] One feasible method is: performing DBF in the azimuth dimension on the candidate target detection point, performing constant false alarm processing in the azimuth dimension in combination with the noise floor estimation result, and obtaining a target screening result in the azimuth dimension; and / or performing DBF in the elevation dimension on the candidate target detection point, performing constant false alarm processing in the elevation dimension in combination with the noise floor estimation result, and obtaining a target screening result in the elevation dimension.

[0306] Another feasible method is to perform azimuth-pitch two-dimensional DBF on the candidate target detection points, and perform azimuth-pitch constant false alarm processing based on the noise floor estimation result to obtain azimuth-pitch two-dimensional target screening results.

[0307] The noise floor estimation result may be obtained by performing statistics on the DBF spectrum of the current dimension and selecting a quantile value (such as the median or mean) as the noise floor estimation result of the current dimension.

[0308] For example, the azimuth angle constant false alarm processing, the elevation angle constant false alarm processing, or the azimuth and elevation angle constant false alarm processing can all be performed in the following manner:

[0309] During constant false alarm processing, one or more of the following operations are performed on the candidate target detection points to obtain a target detection result: global maximum screening, first threshold screening, and second threshold screening, wherein the global maximum screening is used to screen whether the global maximum is a target detection point, the first threshold screening determines whether the current candidate screening result is a target detection point based on the relationship between the candidate screening result and the noise floor estimation value, and is used to filter out false targets, and the second threshold screening determines whether the current candidate screening result is a target detection point based on the relationship between the candidate screening result and the global maximum, and is used to achieve multi-target detection.

[0310] Among them, the global maximum screening includes: judging whether the difference between the digital beamforming spectrum amplitude value corresponding to the current global maximum and the noise floor estimation value is within a first preset range; if it is within the first preset range, the candidate screening result corresponding to the global maximum is the target detection point; the first threshold value screening includes: judging whether the difference between the current candidate screening result and the noise floor estimation value is within a second preset range; if it is within the second preset range, the current candidate screening result is the target detection point; the second threshold value screening includes: judging whether the difference between the power value of the current candidate screening result and the power of the global maximum is within a third preset range; if it is within the third preset range, the current candidate screening result is the target detection point.

[0311] In an exemplary embodiment, when detecting a target in a region of interest based on the RD map, the region of interest is pre-divided into multiple target sub-regions, and the number and position of target points in each sub-region are determined based on the target detection point positions obtained after target detection. It is determined whether there is a target to be detected in the current sub-region based on whether the ratio of the number of target points in each sub-region to the total number of targets is greater than a preset ratio value.

[0312] When applied to enclosed or relatively enclosed spaces, the system can be pre-divided to define and detect different areas of interest (divided areas). For example, when detecting targets within a vehicle cabin, the radar monitoring area can be divided into seating areas and aisle areas, for example. Parameter types and thresholds can then be preset for each type of area, and the corresponding processing steps can be combined to achieve precise detection of targets of interest within a specific area.

[0313] In some optional embodiments, for family cars, the interior space can generally be simply divided into headroom, rear space, and trunk space. If the rear space is the key monitoring area, it can be further divided into seat zones and aisle zones. The seat zones can be divided into a corresponding number of seat zones based on the seats. Similarly, the aisle zones can be divided into a corresponding number of aisle zones corresponding to the seat zones. For example, for a five-seater family sedan, the three seats in the rear space can be divided into three seat zones and three aisle zones. For different types of zones (or zones or zones), corresponding parameter types and thresholds can be preset, and adaptive signal data processing methods and steps can be used to achieve accurate detection of the target of interest (specific target) in a specific zone or zone unit. Adjacent zones can have partial overlap, or they can be adjacent or separated by a gap of a preset width.

[0314] The present disclosure also provides a target detection method for detecting a target in a specific target area. The method includes: performing a first constant false alarm processing on range-Doppler data, and then performing a second constant false alarm processing in the angle dimension to obtain target data.

[0315] In the above-mentioned embodiment, the target's speed is determined by using a frame-level FFT instead of the traditional chirp-level Doppler FFT to achieve accurate detection of the target of interest, and the frequency of updating the results can be further increased by performing sliding window FFT processing on multiple frames, thereby improving the real-time performance of the system. In addition, the amount of data to be processed can be reduced by processing only the distance and / or Doppler area of ​​interest. In addition, when estimating the background noise, the background noise is obtained by taking the minimum value based on the estimate of the current range bin combined with the estimate of the last range bin of interest, which can be more suitable for application scenarios in relatively sealed environments such as the interior of a car. When judging the regional logic, by counting the target points detected in each frame according to the preset area design and judging the counting results according to the preset rules, it is possible to accurately judge whether there is a target in each preset area.

[0316] The target detection method disclosed herein is described in detail below through an application example, and the processing process can be shown in FIG32 .

[0317] In step 1, when a target detection system detects a target within its detectable area, the transmitting antenna in the target detection system may send a frequency modulated continuous wave (FMCW) signal containing a number of chirps. The FMCW signal is refracted and / or reflected by the target to form an echo signal, which can be received by the receiving antenna in the target detection system.

[0318] The target detection system may be an antenna array system or radar system having at least one transmit and receive channel, such as a Multiple Input Multiple Output (MIMO) radar system or a Single-Input Multiple Output (SIMO) system. The target detection system typically has N transmit antennas (TX) and M receive antennas (RX), where N and M are positive integers greater than or equal to 1. By properly arranging the positions of each antenna, a virtual antenna array with N×M transmit and receive channels can be formed.

[0319] Step 2: Perform distance dimension FFT (1D-FFT) processing on each chirp data of each frame of echo signal received to obtain distance dimension data, i.e., 1D-FFT data;

[0320] Before performing 1D-FFT processing, optionally, the chirp data can be first subjected to analog-to-digital conversion (ADC), and then the obtained ADC data is subjected to a first DC filtering process. The first DC filtering process includes: averaging the data collected by each RX channel of each chirp along the fast time dimension, and subtracting the DC component of all sampling points of each RX channel, where the fast time dimension refers to a data sequence formed by sampling the radar echo signal in the distance (or angle) direction. DC filtering can eliminate the DC offset in the signal and make the center line of the signal zero.

[0321] Optionally, before 1D-FFT processing, the DC-filtered data can be windowed, and then the 1D-FFT processing can be performed to obtain distance dimension information. Windowing can reduce spectral leakage and improve spectrum estimation performance. Windowing involves multiplying the original signal by a window function, which can be a weighted function with a specific shape, such as 64, 96, or 128. Common window functions include rectangular windows, Hamming windows, Hanning windows, and Blackman windows.

[0322] The 1D-FFT data can be moved from the BB to the SRAM cache in the CPU or MPU via DMA.

[0323] Step 3: Perform velocity-dimensional FFT (2D-FFT) processing on the 1D-FFT data of the preset number of frames to obtain distance-velocity dimension data, namely 2D-FFT data (or RD graph).

[0324] The velocity dimension can also be called the Doppler dimension, and the above-mentioned distance dimension-velocity dimension information can also be called distance dimension-Doppler information.

[0325] This step differs from the slow-time processing between chirps used in radar signal processing in related technologies. By accumulating multiple frames of data and performing inter-frame 2D-FFT processing on a preset number of 1D-FFT frames using a sliding window, the update frequency of the results can be increased, thereby improving the real-time performance of the system. The 2D-FFT processing in this step can also be called inter-frame FFT processing. For example, with a preset number of frames of 128 and a sliding window step of 1, the first 2D-FFT processing is performed on frames 0-127, and the second 2D-FFT processing is performed on frames 1-128. A sliding window step of 1 is used as an example here; the sliding window step can be configured to a positive integer greater than or equal to 1, and the preset number of frames can also be configured.

[0326] Before performing 2D-FFT processing, the 1D-FFT data can optionally undergo a second DC filtering process. This second DC filtering process involves calculating the complex average along the frame dimension of a preset number of 1D-FFT data frames, obtaining the average value for different channels and distances as its DC component, and subtracting the DC component from each range bin (range bin) in each channel. A range bin is a small distance interval divided by the radar system along the radial direction (i.e., the direction of transmitted signals and received echoes). A range bin represents the discrete distance interval used by the radar system when detecting and tracking targets.

[0327] Optionally, before 2D-FFT processing, the data after the second DC filtering can be windowed, and then 2D-FFT processing can be performed along the frame dimension to obtain range-velocity dimension data. Windowing can reduce spectral leakage and improve the performance of spectrum estimation. The windowing operation involves multiplying the original signal by a window function, which can be a weight function with a specific shape, such as a 64, 96, or 128-degree window function.

[0328] For each chirp data of each frame, when the data cached in the SRAM reaches a certain number of frames, such as 128 frames, the 128 frames of data can be moved from the SRAM to the BB via DMA, and the following steps are continued.

[0329] Step 4, performing non-coherent integration (NCI) on the multi-channel 2D-FFT data to obtain accumulated data;

[0330] Incoherent accumulation can be performed using any of the following formulas:

[0331] Where P(r,v) is the power at the rth range unit and the vth Doppler unit, A(r,v) is the amplitude at the rth range unit and the vth Doppler unit, and s(r,t,a,v) is the complex value of the echo signal at the tth transmit channel and the ath receive channel of the rth range unit and the vth Doppler unit after the above 2D-FFT processing.

[0332] After incoherent accumulation, the result shown in Figure 33 is obtained. The horizontal axis in the figure describes the Doppler, that is, the speed, and the vertical axis describes the distance. The 0th distance unit includes all Doppler values ​​with the vertical axis of 0 in the figure.

[0333] Step 5: Estimate the noise floor, i.e. background noise, of each distance unit in the accumulated data;

[0334] In this example, the following noise floor estimation method is used: n i ′=min(n i ,α*n g), where n i is the original noise floor estimate of the i-th distance unit, and n g is the mean of the noise floor estimates of multiple range cells, or the noise floor estimate of the last range cell of interest, n i ′ is the adjusted noise floor estimate for the i-th range bin, with α > 1. As shown in Figure 34, each grid represents the raw noise floor estimate for a range bin. For example, the median or mean of each range bin can be calculated as the raw noise floor estimate. For example, the noise floor estimate for multiple range bins can be the median of the multiple raw noise floor estimates shown in the dashed box ① in the figure. Using this noise floor estimate as the global noise floor estimate can help adjust the noise floor (also known as noise floor saturation processing). When a target (such as an adult) moves significantly, the Doppler spectrum value will be biased upward, resulting in an overly high raw noise floor estimate, which can lead to undetectable targets during subsequent target screening. Adjusting the noise floor using the global noise floor estimate can make subsequent target detection more accurate. The number and position of the range bins used to calculate the global noise floor estimate are configurable. For example, the noise floor estimates for the multiple range bins shown in the dashed box ② in the figure can be used as the global noise floor estimate.

[0335] Step 6: Perform NR-CFAR (Non-Coherent Constant False Alarm Rate) to obtain candidate target detection points.

[0336] Step 7: Perform the first target screening on the candidate target detection points;

[0337] For each candidate target detection point T i The following operations are performed: extract the 2D-FFT data corresponding to the detection point, select the azimuth antenna and perform azimuth DBF (digital beamforming) according to the azimuth steering vector to obtain the DBF spectrum. The DBF spectrum shows how the signal power or intensity in different directions changes with frequency. The noise floor is estimated on the DBF spectrum and azimuth CFAR (also known as Az-CFAR (Azimuth CFAR)) is performed. The results of the azimuth CFAR are judged under preset conditions. The candidate target points that meet the preset conditions are used as the first target screening results to form the first candidate target point set, which includes all candidate target points T that meet the preset conditions. i,j and its corresponding azimuth angle θ i,j .

[0338] Figure 3 shows the azimuth DBF spectrum statistics (the elevation CFAR results are similar). The x-axis represents azimuth, and the y-axis represents DBF spectrum amplitude. If only the DBF amplitude peak is used as the target point, both the global max and local max values ​​in the figure will be used as target points, which may lead to false alarms. Therefore, to suppress false targets, perform the following operations:

[0339] Perform statistics on the azimuth DBF spectrum and select a quantile value (such as the median or mean) as the noise floor estimate of the azimuth DBF, which is Az_HIST as shown in the figure, denoted as n a ; Find the global maximum in the DBF spectrum (global max), whose value is recorded as P max , judge if P max <β1·n a , β1>1, then the candidate target point corresponding to the global maximum in the DBF spectrum is determined to be a false target point, and the processing of the target point is terminated. In the formula, β1 is one of the thresholds of the azimuth dimension CFAR. The above formula can be used to determine whether the global maximum is a target detection point; if P max ≥β1·n a , then the following processing is performed:

[0340] Traverse all points on the DBF spectrum. If any point satisfies the following conditions, it is determined to be a candidate target point that meets the preset conditions: i ≥β2·n a Formula 1 P i ≥γ·P max Formula 2 P i ≥ρ1·P i-1 Formula 3 P i ≥ρ2·P i+1 Formula 4

[0341] Where i is the serial number of the candidate target point in the orientation dimension, 0 <i<N a -1, N a is the number of points in the azimuthal DBF spectrum, P iis the power of the candidate target point, β2, γ, ρ1, and ρ2 are configurable parameters, 0<γ<1, β2>1. Equation 1 above indicates that the ratio of the power of the candidate target point to the noise floor must be above the preset threshold β2, which is equivalent to setting a first threshold. The first threshold is the noise floor estimate + the first threshold. By setting the first threshold, false target points can be filtered out. Equation 2 indicates that the ratio of the power of the candidate target point to the global maximum power must be above the preset threshold γ, which is equivalent to setting a second threshold. The second threshold is the global maximum power - the second threshold. By setting the second threshold, multi-target detection can be achieved, which is beneficial for identifying targets such as children or infants. When ρ1=ρ2=1, Equations 3 and 4 above indicate that the point is a local maximum.

[0342] Step 8, conduct a second target screening;

[0343] For each candidate target point T i,j According to its azimuth angle θ i,j And the array arrangement generates its pitch direction steering vector sv(i,j) respectively, for each candidate target point T i,j Perform pitch-dimensional DBF to obtain its pitch-dimensional DBF spectrum, estimate the noise floor of its pitch-dimensional DBF spectrum and perform secondary CFAR (called El-CFAR), and perform preset condition judgment on the result of pitch-dimensional CFAR. The candidate target points that meet the preset conditions are used as the second target screening results to form the second candidate target point set, which includes all candidate target points T that meet the conditions. i,j,k and its corresponding pitch angle

[0344] The second target screening process is similar to the first target screening process. First, the pitch DBF is statistically analyzed to obtain the pitch dimension CFAR result. The result is first judged and the formula P is used to calculate the pitch dimension CFAR result. max <β1·n a Determine whether the global maximum is the target detection point. If P max ≥β1·n a , then all points on the DBF spectrum are traversed. For any point that satisfies the above formulas 1-4, it is determined to be a candidate target point that meets the preset conditions.

[0345] The embodiment of the present disclosure performs secondary CFAR on the azimuth and elevation DBFs after performing NR-CFAR, thereby suppressing false targets and achieving, to a certain extent, resolution of multiple targets at the same distance and speed.

[0346] When generating the DBF spectrum of the pitch dimension in the above-mentioned El-CFAR, the pitch dimension steering vector is regenerated according to the azimuth information obtained by Az-CFAR. In an exemplary embodiment, the pitch dimension steering vector with a preset azimuth angle (such as 0° azimuth angle) can also be directly used.

[0347] The order of the first target screening in step 7 and the second target screening in step 8 can be exchanged. In some embodiments, only step 7 or only step 8 can be performed.

[0348] Step 9, target information extraction;

[0349] For all candidate target points T that meet the conditions i,j,k Extracted information includes, but is not limited to, the range bin index (range bin index) and Doppler bin index of the target point. Furthermore, it may include the SNR (signal-to-noise ratio) of the RD-CFAR (Range-Doppler CFAR), the SNR of the Az-CFAR, and the SNR of the El-CFAR (Elevation CFAR). These SNRs can be used for subsequent target personnel determination. For example, weights can be assigned to corresponding target detection points based on the SNR values, where the weights are related to the SNR values.

[0350] Step 10: Count all target points detected in each frame according to the preset area design, and judge the counting results according to the preset rules to obtain whether there is a person in the processing and determine the location of the person, that is, obtain the target information.

[0351] One implementation of the preset area is shown in the figure below, where the horizontal axis is the X direction (vehicle width direction) and the Z direction (vehicle tail-to-vehicle front direction). The interior space is divided into areas that need to be determined along all or some of the XYZ dimensions, or along all or some of the distance-azimuth-pitch dimensions of the polar coordinate system. As shown in Figure 35, an example is given, which divides the three rear seats and the three aisles in front of the three seats into six areas. These areas may overlap with each other or not; some areas in the car may belong to multiple preset areas at the same time, or may not belong to any area.

[0352] Assume that a total of 11 target points are detected in a certain process, and the number of these target points in these preset areas is counted. In this example, the counts of these target points in the areas seat A, seat B, seat C, aisle A, aisle B and aisle C are: 0, 2, 6, 0, 1 and 0.

[0353] Taking the above partitioning and counting results as an example, determining whether there is a person in the process and determining the person's location includes the following steps:

[0354] Step 10.1 Initialize the flags flag_region_i of the six regions to 0, i.e., no one exists (i = 0, 1, ..., 5);

[0355] Step 10.2, determine whether the total valid target number total_valid_tgt_num is 0. If so, determine that there is no one in the car and end. If not, execute step 10.3;

[0356] Step 10.3, traverse all regions, and for region i, determine if the ratio of the statistical value of the number of valid targets belonging to the region tgt_num_region_i to the total number of valid targets total_valid_tgt_num exceeds a preset ratio value (for example, 25%, which can be set based on experience), then set the flag flag_region_i of region i to 1;

[0357] Step 10.4, if there is a region, if its flag_region_i is 1, then set flag_region_empty to 0, that is, there is someone in the car;

[0358] Step 10.5, traverse the seat regions. If flag_region_i is 1 in any seat region, determine that no one is in the corresponding aisle region (for example, if seat A is occupied, then determine that no one is in aisle A).

[0359] Step 10.6, output the results.

[0360] By dividing the area and setting the preset ratio value, the target person can be identified more accurately to prevent misjudgment.

[0361] In an exemplary embodiment, based on the first CFAR result (i.e., the result of step 6), a two-dimensional DBF spectrum in azimuth and elevation can be directly performed, and a second CFAR can be directly performed on the two-dimensional DBF spectrum in azimuth and elevation. The main steps include:

[0362] Extract 2D-FFT data, select antennas and perform two-dimensional DBF according to the azimuth and elevation two-dimensional steering vectors to obtain their DBF spectrum. Estimate the noise floor of the DBF spectrum. The method for estimating the noise floor is as described in step 5 above. Perform secondary CFAR (called AE-CFAR) to obtain all candidate target points T that meet the conditions. i,j and its corresponding azimuth angle θ i,j and pitch angle One possible approach is to traverse all points on the DBF spectrum and determine that any point is a candidate target point if it satisfies the following conditions: i,j ≥β2·n a Formula 5 P i,j ≥γ·P max Formula 6 P i,j ≥ρ1·P i-1,j Formula 7 Pi,j ≥ρ2·P i+1,j Formula 8 P i,j ≥ρ3·P i,j-1 Formula 9 P i,j ≥ρ4·P i,j+1 Formula 10

[0363] Where i is the serial number of the point in the azimuth dimension, 0 <i<N a -1, N a is the number of points in the azimuth dimension DBF spectrum, j is the serial number of the point in the pitch dimension, 0 <j<N e -1, N e is the number of points in the pitch-dimensional DBF spectrum, P i,j is the power of the point, β2, γ, ρ1, ρ2, ρ3, and ρ4 are configurable parameters, 0<γ<1, β2>1. Equation 5 above indicates that the ratio of the power of the candidate target point to the noise floor must be above the preset threshold β2, which is equivalent to setting a first threshold. The first threshold is the noise floor estimate + the first threshold. By setting the first threshold, false target points can be filtered out. Equation 6 indicates that the ratio of the power of the candidate target point to the global maximum power must be above the preset threshold γ, which is equivalent to setting a second threshold. The second threshold is the global maximum power - the second threshold. By setting the second threshold, multi-target detection can be achieved, which is beneficial for identifying targets such as children or infants. When ρ1=ρ2=ρ3=ρ4=1, Equations 7-10 above indicate that the point is a local maximum.

[0364] Figures 36A-36D show the processing results in a single-person scene (baby in aisle C). There are 500 frames of data in total, and 128 frames of inter-frame FFT are performed. The sliding window length of the sliding window processing is 1, and a total of 373 times are processed. Figures 36A and 36C are the results of using only NR-CFAR in the RD dimension and then performing azimuth pitch DoA (angle of arrival estimation); Figures 36B and 36D are the processing results of using NR-CFAR in the RD dimension and then performing azimuth pitch DBF and performing secondary CFAR; Figures 36A and 36B are the pitch-azimuth results, and Figures 36C and 36D are the results of azimuth-processing sequence number. It can be seen that the method of this embodiment (using NR-CFAR in the RD dimension and then performing azimuth pitch DBF and performing secondary CFAR) can effectively reduce the number of invalid false target points, make the target points more concentrated, and effectively reduce the difficulty of post-processing and improve the detection effect.

[0365] Figures 37A-37D show the processing results in a two-person scenario (the baby is in seat B and the adult is in seat A). Figures 37A and 37C are the results of using only NR-CFAR in the RD dimension followed by azimuth and pitch DoA; Figures 37B and 37D are the processing results of using NR-CFAR in the RD dimension followed by azimuth and pitch DBF and secondary CFAR; Figures 37A and 37B are the pitch-azimuth results, and Figures 37C and 37D are the results of azimuth-processing sequence number. It can be seen that the method of this embodiment can effectively reduce the number of invalid false target points, make the target points more concentrated, and effectively reduce the difficulty of post-processing and improve the detection effect.

[0366] The embodiments of the present disclosure are not only applicable to human target detection and positioning scenarios in vehicles, but also to other similar application scenarios such as indoor personnel detection and factory personnel detection.

[0367] The above operations can be performed in the MCU or in the baseband accelerator, or partially in the MCU and partially in the baseband accelerator. For example, 1D-FFT, 2D-FFT, CFAR, and DBF are performed in the baseband accelerator, while target detection and position determination are performed in the MCU.

[0368] Compared with existing technologies, the CPD processing method proposed in the embodiments of the present disclosure mainly adopts a new inter-frame accumulation method, performs coherent accumulation at the frequency of human breathing, improves the signal-to-clutter ratio of human targets and static strong clutter, and thus improves detection performance. In addition, CFAR secondary detection is again performed on the DBF spectrum generated by the CFAR result in the azimuth / pitch dimension, effectively eliminating false alarms and effectively improving detection and measurement performance. As a result, special targets or weak targets such as infants and young children in the cabin can be accurately detected, realizing applications such as child in-vehicle detection (CPD) and seat belt reminder devices (SBR).

[0369] Performance evaluation was carried out through a large number of actual experiments, and the results showed that the method proposed in the present disclosure can achieve a detection rate of more than 99%, a missed detection rate and a false alarm rate of less than 0.5% and 1%, respectively, which has obvious advantages over existing technologies.

[0370] The present disclosure also provides a method for target detection in a specific target area. The method includes performing a first constant false alarm (CFAR) process on range-Doppler data and then performing a second CFAR process on the angle dimension to obtain target data. The specific target area may be, for example, an enclosed and / or semi-enclosed area such as a cabin or a room. The first and second CFAR processes may be described in the aforementioned embodiments.

[0371] Figure 38 is a schematic flow chart of post-processing for a target in an embodiment of the present application. As shown in Figure 38, on the basis of the process structure shown in Figure 1, after performing azimuth / pitch angle estimation, the target point cloud data output can be processed in combination with a machine learning model to accurately detect the target in the area of ​​interest. For example, for the single target point cloud data obtained after azimuth / pitch angle estimation, the false alarm suppression technology of ML (Machine Learning, such as SVM or RF algorithm) can be combined to suppress non-ideal data such as false alarms generated by noise, stationary clutter or target multipath, that is, for target point cloud data such as pitch DBF&DoA output, operations such as ML-based false-alarm suppression, clustering and / or ML-based target classification can be continued, where the clustering process can be an algorithm such as agglomerative clustering or DBSCAN. At the same time, the clustered point cloud data can also be input into a trained machine learning model (such as SVM or RF) to determine whether there is a person in the current processing. If there is, its location can be further determined, and operations such as distinguishing and judging physical targets such as adults, children, and infants can be realized.

[0372] In the embodiment shown in FIG38 , the influence of non-ideal factors such as noise, stationary clutter and target multipath can be effectively reduced, and the difficulty of selecting regional parameters and the difficulty of designing complex regional judgment logic can be effectively reduced, thereby more effectively utilizing point cloud information, especially for applications in complex scenarios, and achieving better target detection and discrimination performance.

[0373] FIG39 is a schematic diagram of a process for post-processing a target in conjunction with a DL algorithm in accordance with an embodiment of the present application. As shown in FIG39 , based on the process shown in FIG1 , after extracting data obtained from at least one of the following steps: multi-frame accumulation (store frame), frame Fourier transform (Frame FFT), non-coherent integration, and constant false alarm rate (CFAR), a deep learning (DL) algorithm can be used to determine, locate, and identify the target, such as identifying whether the physical target is an adult or a child (Adult / Child Classification).

[0374] For example, the raw data of 1D-FFT can be used as the input of DL (Access 1D-FFT data as raw input of DL), the raw data of 2D-FFT between frames can be used as the input of DL (Access 2D-FFT data as raw input of DL), the non-coherent integrated 2D-FFT data can be used as the input of DL (Access non-coherent integrated 2DFFT data as raw input of DL), and / or the SNR data can be used as the input of DL (Access SNR data as raw input of DL) to achieve the distinction and judgment of subsequent targets. The raw data here can be complex data, the modulus value of complex data, the modulus square of complex data, or other transformation forms, and can be values ​​in the linear domain, values ​​in the dB (log) domain, or other calculation forms.

[0375] Based on the process shown in Figure 39, by adopting at least one of the above methods to extract multi-frame 1D-FFT data, multi-channel data after multi-frame FFT, and energy (power) or SNR data after multi-frame FFT and SISO-combine, the extracted data is input into the constructed deep learning model (such as CNN or Transformer, etc.) to determine whether there is a target in the current processing, and to determine the current target type (such as determining whether the target is an adult or a child, etc.).

[0376] It should be noted that the process steps and related descriptions shown in Figure 39 are compatible with the process steps and related descriptions shown in Figure 38; for example, the Region HIST and its subsequent modules in the process shown in Figure 39 can be partially or completely replaced by the ML-based false alarm suppression and its subsequent modules in the process shown in Figure 38.

[0377] Figure 40 is a flow chart of the target detection method combined with the DL algorithm in an embodiment of the present application. As shown in Figure 40, on the basis of the process shown in Figure 1, after the inter-frame FFT (Frame FFT) processing, two-dimensional digital beamforming (2D-DBF) processing and DL-based target classification processing (DL-based Classification) operation can be directly performed. That is, after the inter-frame FFT, CFAR processing is not performed, but 2D-DBF operation is performed on the azimuth and pitch dimensions of each region of interest (or unit interval) to obtain a number of RD maps corresponding to the region of interest; subsequently, the above-mentioned RD maps are input into the constructed deep learning model (DL) for target judgment. In some optional embodiments, the relationship between the above-mentioned region of interest and the RD map can also be one-to-many, that is, at least two RD maps can be obtained based on one region of interest, and the detailed number can be adjusted according to actual needs.

[0378] The DL input can be linear or dB-domain amplitude or power, and can also be normalized. The deep learning model constructed above can also classify the input, and the categories distinguished can include: whether there is a target, the attributes of the target, and the specific unit interval location (i.e., regional location) of each target.

[0379] For example, in an application scenario where the rear area of ​​a car is the area of ​​interest or the target area, and is divided into three seating areas (interval units) and corresponding three aisle areas, after the inter-frame FFT, a 2D-DBF of azimuth and pitch can be performed on the center positions of the above-mentioned six areas of interest (the areas of interest are also set to three seating areas) to obtain six RD maps (or three RD maps) corresponding to the areas of interest. At least part of the above-mentioned six RD maps (or three RD maps) are input into the constructed deep learning model for operations such as target judgment and differentiation. For example, the constructed deep learning model performs classification based on the input to determine whether there is a living target in the above-mentioned area of ​​interest. If there is a living target, it can further determine whether the living target is an adult, child, infant, or pet, etc., and can further determine the specific location information of the living target, such as which seat or aisle area it is in.

[0380] In the embodiment shown in FIG40 , the target detection method is a processing flow based on dual drive of model data, which has low requirements for signal processing and can effectively avoid steps such as the selection of signal processing parameters, regional parameters, and the design of CFAR logic and regional judgment logic, thereby greatly reducing the difficulty of solution implementation and design.

[0381] In some optional embodiments, in all the aforementioned target detection methods, all FFT processing may be windowed FFT, and may also be implemented by using SVA combined with FFT.

[0382] FIG41 is a flow chart of another target detection method incorporating a DL algorithm in an embodiment of the present application. As shown in FIG41 , based on the process shown in FIG40 , after the range-dimensional Fourier transform (Range FFT, also known as 1D-FFT), a 2D-DBF can be performed first, followed by frame data accumulation and inter-frame FFT processing. That is, before the inter-frame FFT, a 2D-DBF of the azimuth and elevation angles of the center position of the region of interest (e.g., a three-seat area) is first performed. This embodiment enables parallel processing of data processing and wave transmission time, thereby effectively reducing processing time.

[0383] In some optional embodiments, in all the aforementioned target detection methods, the multi-frame sliding window FFT can be replaced by other time-frequency transform processing, such as short-time Fourier transform, fractional Fourier transform, etc., which can also achieve effective detection of the target of interest.

[0384] In some optional embodiments, in all the aforementioned target detection methods, the multi-frame sliding window FFT can also be replaced by a high-pass FIR, a Comb FIR, or a filter optimized and synthesized according to an optimization function, so as to achieve the purpose of obtaining similar or even higher performance of multi-frame FFT processing with fewer hardware resources.

[0385] As shown in Figures 40 and 41, the combination of 2D-FFT and DL-based Classification can effectively improve the accuracy of target recognition. At the same time, the 2D-FFT steps and DL recognition steps can be flexibly set between various steps in signal processing according to actual needs. For example, the 2D-FFT steps can be set after the Range FFT or after the Frame FFT.

[0386] Figure 42 is a flow chart of another target detection method combined with a DL algorithm in an embodiment of the present application. As shown in Figure 42, on the basis of the process shown in Figure 40 or Figure 41, that is, after the range-dimensional Fourier transform (Range FFT, also known as 1D-FFT), the accumulation of frame data can be completed after continuing 2D-DBF, and the target classification operation based on DL (Complex DL-based Classification) can be directly performed. That is, in the entire target detection process, no inter-frame FFT processing operation is performed. After 1D-FFT, 2D-DBF is directly performed on the center position of the area of ​​interest (such as 3 seat areas) to obtain the corresponding number of distance-frame maps (such as 3 range-frames), and the above-mentioned distance-frame maps are subsequently input into the constructed (for example, complex) deep learning model to perform operations such as target differentiation and judgment.

[0387] In the embodiment shown in FIG42 , since the FFT processing step between frames is avoided, the amount of calculation can be effectively reduced; at the same time, if a complex deep learning model is adopted, a higher processing gain can be obtained compared to FFT, thereby achieving the purpose of improving target detection performance; in addition, this embodiment also effectively shortens the processing level, which can facilitate its efficient scheduling operations, etc.

[0388] In some optional embodiments, the DBF in the target detection method shown in Figures 40-42 can be replaced by Capon, MUSIC, ESPRINT, etc. and their variant algorithms. At the same time, the beam synthesis at the center of the area of ​​interest and / or the antenna arrangement can be optimized and integrated to further improve the system target detection performance.

[0389] It should be noted that the embodiments of the present application can refer to each other and be compatible when there is no conflict, and the order and whether to set up each functional module can also be adjusted according to needs. At the same time, for a system with a BB module and an MCU module, when the system detects a target by emitting electromagnetic waves and receiving corresponding echo signals, each step in the target detection method of the embodiments of the present application can be configured to run in the BB module and / or MCU module according to actual needs and the data processing capacity and timeliness of the system operation. The relevant examples in the figure can be used as a reference for some of the options.

[0390] In some optional embodiments, all of the aforementioned target detection methods are applicable to TDM-MIMO systems. For systems that do not adopt TDM-MIMO, target detection and judgment operations can be performed based on presetting different phase shifts during different Tx transmissions, and after directly transmitting waveform digital synthesis (phased array) of the region of interest, the processing flow shown in Figures 40-42 can be combined.

[0391] In other embodiments, for systems employing pulse, pulse compression, or ultra-wideband (UWB) technologies, target detection can be performed directly using the method flows shown in Figures 40-42. Transmit beam digital synthesis technology can also be employed to further enhance target detection performance. For pulse compression systems, 1D-FFT processing may be unnecessary, but a pulse compression step is required.

[0392] It should be noted that the above-mentioned processing methods corresponding to systems of different standards can also be applied to systems of mixed standards. For example, for a system with 6 transmitting antennas (i.e., 6 Tx), 3 of the transmitting antennas (3 Tx) can be TDM standards, and the other 3 transmitting antennas can be other standards. In subsequent processing, the processing method corresponding to the above-mentioned standards can be used.

[0393] After the inter-frame FFT (Frame FFT) processing, two-dimensional digital beamforming (2D-DBF) processing and DL-based target classification processing (DL-based Classification) operations are directly performed. That is, after the inter-frame FFT, CFAR processing is not performed, but 2D-DBF operations are performed on the azimuth and elevation angles of each region of interest (or unit interval) to obtain a number of RD maps corresponding to the region of interest; the above-mentioned RD maps are subsequently input into the constructed deep learning model (DL) for target judgment. In some optional embodiments, if the inter-frame FFT method is not used, but FIR, STFT (Short-Time Fourier Transform) and other methods are used for processing, the above-mentioned RD map is a one-dimensional map output corresponding to FIR, and a three-dimensional map output corresponding to STFT time-frequency transform.

[0394] It should be noted that the electromagnetic waves in this application may include radio waves and light waves. Radio waves include short waves, medium waves, long waves and microwaves, etc. Microwaves include centimeter waves (i.e. electromagnetic waves in the range of 3GHz to 30GHz, such as electromagnetic waves in the range of 3.1GHz to 10.6GHz, electromagnetic waves in the 24GHz frequency band, etc.) and millimeter waves (i.e. electromagnetic waves in the range of 30GHz to 300GHz, such as electromagnetic waves in the 60GHz frequency band, electromagnetic waves in the 77GHz frequency band (such as 77GHz to 81GHz, etc.)). Light waves may include ultraviolet rays, visible light, infrared rays, and lasers, etc., among which the electromagnetic wave frequency band of lasers is (3.846 to 7.895)*10^5GHz, that is, lasers are included in part of the frequency bands of ultraviolet rays and visible light.

[0395] An embodiment of the present disclosure also provides an integrated circuit, which may include: a signal transmitting module, configured to transmit electromagnetic waves for target detection; a signal receiving module, configured to receive echoes formed by the reflection and / or scattering of the electromagnetic waves; and a processing module, configured to perform signal and data processing on the echoes to achieve target detection.

[0396] Optionally, the target detection is, for example, at least one of the following operations on the target in the region of interest: judgment, positioning and / or identification.

[0397] In an exemplary embodiment, the processing module includes at least a baseband unit, and the baseband unit can be configured to implement the distance dimension FFT processing, velocity dimension FFT, first constant false alarm processing, and second constant false alarm processing in the method as described in any embodiment of the present disclosure.

[0398] Optionally, in an exemplary embodiment, the processing module may further include an MCU unit, and when the method includes digital beamforming and frame data accumulation, the baseband unit is configured to implement the digital beamforming, and the MCU unit is configured to implement the frame data accumulation.

[0399] In an exemplary embodiment, the integrated circuit may be a millimeter wave radar chip or die.

[0400] The present application also provides an integrated circuit, which may include a radio frequency module, an analog signal processing module, and a digital signal processing module connected in sequence; the radio frequency module is used to generate a radio frequency transmission signal and receive an echo signal; the analog signal processing module is used to down-convert the echo signal to obtain an intermediate frequency signal; the digital processing module is used to perform analog-to-digital conversion on the intermediate frequency signal to obtain a digital signal; and the digital signal is processed based on the target detection method in the embodiment of the present application to achieve the purpose of target detection. For example, the integrated circuit may be a millimeter wave radar chip (chip or die). The digital processing module may include subunits such as a BB unit and an MCU unit, and each subunit may be configured to execute each corresponding step in the target detection method in the above embodiment.

[0401] In some optional embodiments, the integrated circuit may be an AiP (Antenna-In-Package) chip structure, an AoP (Antenna-On-Package) chip structure, or an AoC (Antenna-On-Chip) chip structure.

[0402] According to other embodiments of the present application, an electromagnetic wave sensor is also proposed. The electromagnetic wave sensor may include an antenna and an integrated circuit as described above. The integrated circuit is electrically connected to the antenna for transmitting and receiving electromagnetic wave signals. For example, the electromagnetic wave sensor may include: a carrier, an integrated circuit and an antenna as described in any of the above embodiments, etc. The integrated circuit may be provided on the carrier; the antenna may be provided on the carrier, or integrated with the integrated circuit as an integral device provided on the carrier (i.e., the antenna may be an antenna provided in an AiP, AoP or AoC structure in this case); the integrated circuit is connected to the antenna (i.e., the sensor chip or integrated circuit does not have an integrated antenna in this case, such as a conventional SoC, etc.) for transmitting and receiving electromagnetic wave signals. The carrier may be a printed circuit board PCB, and the corresponding transmission line may be a PCB trace.

[0403] An embodiment of the present application provides a terminal device, which may include: a device body; and an electromagnetic wave sensor as described above, which is arranged on the device body; wherein the electromagnetic wave sensor is used for target detection and / or communication to provide reference information for the operation of the device body.

[0404] The embodiment of the present application also provides an electronic device (which can be understood as a terminal device), which can be expressed in the form of a general-purpose computing device. The components of the electronic device may include, but are not limited to: at least one processing unit, at least one storage unit, a bus connecting different system components (including a storage unit and a processing unit), a display unit, etc. Among them, the storage unit stores program code, and the program code can be executed by the processing unit so that the processing unit executes the method described in this specification according to various exemplary embodiments of the present application. The storage unit may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) and / or a cache memory unit, and may further include a read-only memory unit (ROM).

[0405] The storage unit may also include a program / utility having a set (at least one) of program modules, such program modules including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0406] The bus can represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0407] The electronic device may also communicate with one or more external devices (e.g., keyboards, pointing devices, Bluetooth devices, etc.), one or more devices that enable a user to interact with the electronic device, and / or any device that enables the electronic device to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface. Furthermore, the electronic device may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter. The network adapter may communicate with other modules of the electronic device via a bus. It should be understood that, although not shown in the figures, other hardware and / or software modules may be used in conjunction with the electronic device, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0408] For example, the electronic device in the embodiment of the present application may further include: a device body; and an electromagnetic wave sensor as described in any of the above embodiments, which is arranged on the device body; wherein the electromagnetic wave sensor can be used to realize functions such as target detection and / or wireless communication.

[0409] Specifically, based on the above embodiment, in an optional embodiment of the present application, the electromagnetic wave sensor can be disposed outside the device body or inside the device body, and in other optional embodiments of the present application, the electromagnetic wave sensor can be disposed partially inside the device body and partially outside the device body. This embodiment of the present application is not limited to this, and the specific configuration may vary depending on the circumstances.

[0410] In an optional embodiment, the above-mentioned device body can be a component and product used in fields such as smart cities, smart homes, transportation, smart homes, consumer electronics, security monitoring, industrial automation, cabin detection (such as smart cockpits), medical equipment and health care. For example, the device body can be intelligent transportation equipment (such as cars, bicycles, motorcycles, ships, subways, trains, etc.), security equipment (such as cameras), liquid level / flow rate detection equipment, smart wearable devices (such as bracelets, glasses, etc.), smart home devices (such as sweeping robots, door locks, televisions, air conditioners, smart lights, etc.), various communication devices (such as mobile phones, tablets, etc.), as well as gates, smart traffic lights, smart signs, traffic cameras and various industrial robotic arms (or robots), etc. It can also be various instruments for detecting life characteristic parameters and various devices equipped with the instruments, such as life characteristic detection in car cabins, indoor personnel monitoring, smart medical equipment, consumer electronic devices, etc.

[0411] An embodiment of the present application further provides a non-transitory computer-readable storage medium having computer-readable instructions stored thereon. When the instructions are executed by a processor, the processor executes the feeder unequal length compensation method as described above.

[0412] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software or by combining software with necessary hardware. The technical solution according to the embodiment of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, or a network device, etc.) to execute the above method according to the embodiment of the present application.

[0413] The software product can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, but not limited to, a system, device or component of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0414] Computer-readable storage media may include a data signal propagated in baseband or as part of a carrier wave, which carries readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination thereof.

[0415] The program code for performing the operations of the present application can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, and the like, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, as a stand-alone software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0416] The computer-readable medium carries one or more programs. When the one or more programs are executed by the device, the computer-readable medium implements the aforementioned functions.

[0417] Those skilled in the art will appreciate that the modules described above can be distributed in the device according to the description of the embodiment, or can be modified accordingly to be used in one or more devices that are different from the embodiment. The modules of the above embodiment can be combined into one module or further divided into multiple submodules.

[0418] According to an embodiment of the present application, a computer program is proposed, including a computer program or an instruction, which, when executed by a processor, can execute the method described above. In an optional embodiment, the above-mentioned integrated circuit can be a millimeter-wave radar chip. The type of digital function module in the integrated circuit can be determined according to actual needs. For example, in a millimeter-wave radar chip, the data processing module can be used for functions such as range-dimensional Doppler transform, velocity-dimensional Doppler transform, constant false alarm detection, direction of arrival detection, point cloud processing, etc., to obtain information such as the distance, angle, speed, height, micro-Doppler motion characteristics, shape, size, surface roughness and dielectric properties of the target.

[0419] It should be noted that radio devices can achieve functions such as target detection and / or communication by transmitting and receiving radio signals to provide detection target information and / or communication information to the device body, thereby assisting or even controlling the operation of the device body.

[0420] For example, when the above-mentioned device body is applied to an advanced driver assistance system (i.e., ADAS), the radio device (such as millimeter-wave radar) as an on-board sensor can assist the ADAS system to realize application scenarios such as adaptive cruise control, automatic braking assistance (i.e., AEB), blind spot detection warning (i.e., BSD), assisted lane change warning (i.e., LCA), reversing assistance warning (i.e., RCTA), parking assistance, rear vehicle warning, collision avoidance, pedestrian detection, and in-cabin liveness detection (i.e., CPD).

[0421] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In hardware implementations, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all components may be implemented as software executed by a processor, such as a digital signal processor or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or temporary medium). As is well known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable, and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those skilled in the art that communication media generally embodies computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0422] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0423] The above-described embodiments merely express preferred embodiments of the present invention and the technical principles employed. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It is apparent to those skilled in the art that various changes, readjustments, and substitutions can be made without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments. Without departing from the concept of the present invention, the present invention may also include more other equivalent embodiments. The scope of protection of the present invention's patent is determined by the scope of the appended claims.

Claims

1. A method for target detection, comprising: accumulating frame data of 1D-FFT data, wherein the 1D-FFT data is obtained based on performing distance-dimensional FFT processing on the echo signal; Based on the accumulated frame data, FFT processing is performed between frames to obtain the RD spectrum; as well as The detection of the target in the region of interest is achieved based on the RD map.

2. The target detection method according to claim 1, wherein: The method of performing inter-frame FFT processing based on the accumulated frame data to obtain an RD spectrum, and detecting a target in a region of interest based on the RD spectrum, includes: Target data is extracted from the accumulated frame data, sliding windowing is performed on the target data, and digital signal processing is performed on the windowed data to obtain target information; wherein the length of the sliding window corresponds to a duration of the same order of magnitude as the period of the periodic motion of the target.

3. The target detection method according to claim 2, wherein: The step of extracting target data from the accumulated frame data comprises: Extracting at least part of the data or at least part of the data processed from each frame of data obtained after the distance dimension FFT processing as the target data; or determining the parameters corresponding to each distance unit in each frame of data according to each frame of data obtained after the distance dimension FFT processing to generate the target data; The step of performing digital signal processing on the windowed data to obtain target information includes: Noise floor estimation is performed within a preset range, wherein an upper limit of the preset range is determined according to a noise floor estimation result of a distance unit with the farthest distance; and constant false alarm detection is performed according to the estimated noise floor to obtain target information.

4. The target detection method according to claim 2, wherein: The step of extracting target data from the frame data obtained after the distance dimension FFT processing comprises: Extracting data within a preset distance range from the frame data obtained after distance dimension FFT processing as the target data; and / or, The step of performing digital signal processing on the windowed data includes: Digital signal processing is performed on the data within the preset Doppler range in the windowed data.

5. The target detection method according to claim 2 or 3, when the data is cached in Ring FIFO, the method further comprises: Data is read from the Ring FIFO, and the read data is rearranged.

6. The target detection method according to claim 2, wherein: The sliding windowing of the target data and digital signal processing of the windowed data to obtain target information includes: Windowing the target data according to the sliding windows corresponding to the cycles of different vital sign parameters of the living body, and performing digital signal processing on the windowed data corresponding to the sliding windows of different lengths respectively; The periods of the vital sign parameters corresponding to different sliding windows are different, and the length of each sliding window has a duration of the same order of magnitude as the period of the vital sign parameter.

7. The target detection method according to claim 1 or 2, wherein: The detecting of the target in the region of interest based on the RD map includes: Based on the RD spectrum, the at least two transceiver channels are individually subjected to constant false alarm processing to obtain candidate target data of the at least two transceiver channels; The candidate target data of at least two transceiver channels are processed to obtain final target data.

8. The target detection method according to claim 7, wherein: The method of performing constant false alarm processing on at least two transceiver channels separately based on the RD spectrum includes: Based on the RD spectrum, performing constant false alarm processing on at least two transceiver channels respectively; or According to the RD spectrum, the noise floor of the transceiver channel that is subjected to the CFAR processing is estimated separately to obtain the noise floor estimation result of the single transceiver channel; according to the noise floor estimation result of the single transceiver channel, the CFAR detection is performed separately on the corresponding transceiver channel.

9. The target detection method according to claim 8, wherein: The noise floor estimation result of the single transceiver channel is composed of the noise floor estimation results of each distance unit on the single transceiver channel, and the constant false alarm detection result of the single transceiver channel is composed of the constant false alarm detection results of each Doppler unit of each distance unit on the single transceiver channel; According to the noise floor estimation result of the single transceiver channel, the corresponding transceiver channel is subjected to constant false alarm detection separately, which is realized by the following expression: Wherein, F(c,r,v) is the constant false alarm detection result of the vth Doppler unit of the rth range unit of the cth channel, P(c,r,v) is the echo energy on the vth Doppler unit of the rth range unit of the cth channel, n′(c,r) is the noise floor estimation result of the rth range unit of the cth channel, β is a preset parameter, β≥1.

10. The target detection method according to claim 8, wherein: The step of separately estimating the noise floor of the receiving and transmitting channels that are separately subjected to constant false alarm processing according to the RD spectrum includes: According to the RD spectrum, the noise floor of the receiving and transmitting channels that are subjected to the CFAR processing are estimated separately within a preset range; wherein the upper limit of the preset range is determined according to the noise floor estimation result of the farthest distance unit.

11. The target detection method according to claim 7, wherein: The processing of the candidate target data based on at least two transceiver channels includes: A constant false alarm detection is performed according to the currently acquired candidate target data and a preset threshold to obtain the final target data.

12. The target detection method according to claim 11, wherein: The constant false alarm detection is performed according to the candidate target data currently obtained and the preset threshold value, which is achieved by the following expression: Wherein, F(c,r,v) is the candidate target data of the vth Doppler unit of the rth range unit of the cth channel, N c is the total number of channels, F0 is the preset threshold, and F′(r,v) is the final target data of the vth Doppler unit of the rth range unit.

13. The target detection method according to any one of claims 2 to 12, after acquiring the target information, the method further comprises: According to the target falling into each preset area, verifying the target in each preset area, wherein the preset area is obtained by dividing the detection space; According to the verified target, the detection result of each preset area is output.

14. The target detection method according to claim 13, wherein: The verifying the target in each preset area according to the target falling in each preset area includes: Verifying the target in each preset area according to the proportion of the target falling in each preset area in the detected targets; or The target in each preset area is verified according to the signal-to-noise ratio of the target falling within each preset area.

15. The target detection method according to claim 1, wherein: The detecting of the target in the region of interest based on the RD map includes: Performing a first constant false alarm processing on the RD spectrum to obtain candidate target detection points; A second constant false alarm processing is performed on the candidate target detection point, and target detection is performed based on the second constant false alarm processing result.

16. The target detection method according to claim 15, wherein: The step of performing a first constant false alarm processing on the RD spectrum to obtain a candidate target detection point includes: Incoherent accumulation is performed on the distance and speed dimension data of multiple channels, and a first constant false alarm processing based on noise estimation is performed based on the incoherent accumulation result to obtain candidate target detection points.

17. The target detection method according to claim 15, wherein: The performing a second constant false alarm processing on the candidate target detection point includes: Performing DBF in azimuth dimension on the candidate target detection point, performing azimuth constant false alarm processing in combination with the noise floor estimation result, and obtaining a target screening result in azimuth dimension; and / or performing DBF in elevation dimension on the candidate target detection point, performing pitch angle constant false alarm processing in combination with the noise floor estimation result, and obtaining a target screening result in elevation dimension; Perform azimuth and elevation two-dimensional DBF on the candidate target detection points, perform azimuth and elevation constant false alarm processing based on the noise floor estimation result, and obtain azimuth and elevation two-dimensional target screening results.

18. The target detection method according to claim 17, wherein: During constant false alarm processing, one or more of the following operations are performed on the candidate target detection point to obtain a target detection result: global maximum value screening, first threshold value screening, and second threshold value screening, wherein the global maximum value screening is used to screen whether the global maximum value is a target detection point, the first threshold value screening determines whether the current candidate screening result is a target detection point according to the relationship between the candidate screening result and the noise floor estimation value, and the second threshold value screening determines whether the current candidate screening result is a target detection point according to the relationship between the candidate screening result and the global maximum value.

19. The target detection method according to claim 18, wherein: The global maximum screening includes: determining whether the difference between the digital beamforming spectrum amplitude value corresponding to the current global maximum value and the noise floor estimation value is within a first preset range, and if within the first preset range, the candidate screening result corresponding to the global maximum value is the target detection point; The first threshold value screening includes: determining whether the difference between the current candidate screening result and the noise floor estimation value is within a second preset range, and if it is within the second preset range, the current candidate screening result is the target detection point; The second threshold value screening includes: judging whether the difference between the power value of the current candidate screening result and the power of the global maximum value is within a third preset range; if within the third preset range, the current candidate screening result is the target detection point.

20. The target detection method according to claim 15, wherein: The method of performing incoherent accumulation on the distance and speed dimension data of multiple channels and performing a first constant false alarm processing based on noise estimation based on the incoherent accumulation result comprises: Incoherent accumulation is performed on the distance and speed dimension data of multiple channels, and the noise floor of each distance unit after the incoherent accumulation is estimated to obtain a noise floor estimation value of each distance unit, and incoherent constant false alarm processing is performed using the noise floor estimation value, wherein when estimating the noise floor of each distance unit after the incoherent accumulation, the noise floor of each distance unit is adjusted using the global noise floor.

21. The target detection method according to claim 15, wherein: The performing target detection based on the second constant false alarm processing result includes: The region of interest is pre-divided into multiple target sub-regions, and the number and position of target points in each sub-region are determined based on the position of the target detection points obtained after target detection. Whether there is a target to be detected in the current sub-region is determined based on whether the ratio of the number of target points in each sub-region to the total number of targets is greater than a preset ratio.

22. The target detection method according to claim 1, wherein: For multi-channel application scenarios, the detection of targets in the region of interest based on the RD map includes: After incoherently accumulating the RD spectra of multiple channels, continue with constant false alarm processing based on noise estimation and direction of arrival estimation to achieve target detection; or After the CFAR processing is performed on the RD spectra of multiple channels respectively, the binary integration processing and the direction of arrival estimation in the channel domain are continued, and then any one of the following operations is performed on the target: judgment, positioning and recognition.

23. The method for target detection according to claim 22, wherein: The constant false alarm processing is to perform DAE CFAR processing on the azimuth angle and the elevation angle respectively.

24. The method for target detection according to claim 1, wherein: The detecting of the target in the region of interest based on the RD map includes: The target is then identified based on extracting at least part of the multi-frame accumulated data, the inter-frame Fourier transformed data, the incoherently accumulated data and / or the constant false alarm detection data.

25. An integrated circuit comprising: A signal transmitting module configured to transmit electromagnetic waves for target detection; A signal receiving module, configured to receive an echo formed by reflection and / or scattering of the electromagnetic wave; as well as A processing module is configured to process the echo according to the target detection method according to any one of claims 1-24 to achieve target detection.

26. The integrated circuit of claim 25, wherein: The processing module includes a baseband unit and an MCU unit, wherein the baseband unit is configured to implement the distance dimension FFT processing and inter-frame FFT processing in the method as described in any one of claims 1-24; the MCU unit is configured to implement the frame data accumulation and detection of targets in the area of ​​interest in the method as described in any one of claims 1-24.

27. The integrated circuit of claim 26, wherein: When the method comprises digital beamforming, the baseband unit is configured to implement the digital beamforming.

28. The integrated circuit of claim 25, wherein: The integrated circuit is a millimeter wave chip or a sensor chip.

29. An electromagnetic wave sensor, comprising: Carrier; An integrated circuit as claimed in any one of claims 25 to 28, arranged on a carrier; An antenna is arranged on the carrier, or the antenna and the integrated circuit are integrated into one device and arranged on the carrier; Wherein, the integrated circuit is connected to the antenna and is used for transmitting the electromagnetic wave signal and / or receiving the echo signal.

30. A terminal device, comprising: Equipment body; as well as The electromagnetic wave sensor as claimed in claim 29 is disposed on the device body; The electromagnetic wave sensor is used for target detection and / or communication to provide reference information for the operation of the device body.

31. A non-transitory computer-readable storage medium having computer-readable instructions stored thereon, which, when executed by a processor, causes the processor to execute the target detection method according to any one of claims 1 to 24.

Images