Boundary detection methods, apparatus, storage media and equipment
By acquiring and analyzing multiple frames of raw target data, the sensor installation method is automatically determined and boundary detection is performed, solving the problem of relying on manual labor to obtain the sensor installation position and achieving efficient and low-cost boundary detection.
Patent Information
- Application Number
- CN202210753540.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-06-28
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2042-06-28
AI Technical Summary
In existing technologies, obtaining the sensor installation location requires human intervention, resulting in high labor costs and a cumbersome detection process, which affects the efficiency and accuracy of boundary detection.
By acquiring at least two frames of raw target data, the sensor installation method is determined based on the analysis of this data, and the boundary detection is performed using an adapted boundary detection module to achieve automatic determination of the sensor installation position.
The sensor installation location can be determined without human intervention, simplifying the detection process, improving detection efficiency and reducing costs.
Smart Images

Figure CN117347952B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of boundary detection technology, and in particular to a boundary detection method, apparatus, storage medium and device. Background Technology
[0002] With the development of detection technology, boundary detection within a specific area has been widely used. In order to improve the accuracy of boundary detection, it is generally necessary to first determine the boundary of the specific area.
[0003] Taking a specific space as an indoor space as an example, the locations where sensors used for boundary detection are installed indoors can generally be divided into: front mounting (hereinafter referred to as front mounting), such as the front of a wall-mounted air conditioner; side mounting (hereinafter referred to as side mounting), such as the side of a floor-standing air conditioner; and top mounting (hereinafter referred to as top mounting), such as the top of a ceiling-mounted air conditioner.
[0004] Indoor boundary detection requires prior knowledge of the radar's installation location, and then the boundary detection parameters and / or strategies need to be adjusted accordingly based on the different sensor installation locations. In the existing technology, obtaining the sensor installation location requires human intervention, which results in high labor costs and a cumbersome detection process. Summary of the Invention
[0005] Therefore, it is necessary to provide a boundary detection method, apparatus, storage medium, and device to address the aforementioned technical problems. This application provides a boundary detection method for using sensors positioned within a detection area to perform boundary detection. The method includes:
[0006] Acquire at least two frames of raw target data;
[0007] Based on the analysis of at least two frames of raw target data, the sensor installation method is determined; and
[0008] Based on the sensor installation method, boundary detection is performed on the area to be detected.
[0009] In this application, at least two frames of raw target data correspond to the area to be detected, and can be at least two frames of raw target data that are consecutive in time. For example, the area to be detected can be an indoor area, such as a room or a stadium.
[0010] Optionally, the sensor can be installed on the top, side, or front.
[0011] Optionally, "obtaining the sensor mounting method based on the analysis of at least two frames of raw target data" includes:
[0012] The single-frame judgment step analyzes a single frame of raw target data to determine the sensor's installation method.
[0013] as well as
[0014] The multi-frame judgment step is based on the judgment results of the installation methods of multiple sensors obtained from the original target data of the preset number of frames in the single-frame judgment step. The results are then quantitatively analyzed to finally confirm the installation method of the sensors.
[0015] Optionally, the single-frame determination step includes:
[0016] On the plane of the sensor's coordinate system, obtain the initial boundary model of multiple (e.g., at least three) boundaries of the region to be detected based on the original target data of the current frame;
[0017] For each initial boundary model, multiple boundary searches are performed within preset angle and distance ranges to determine the boundary parameters of each boundary in the region to be detected. These boundary parameters include confidence levels; and...
[0018] The boundary with the highest confidence level is selected as the feature boundary, and the installation method of the sensor is confirmed based on the correspondence between the angle of the feature boundary and the preset angle range.
[0019] Optionally, the multi-frame determination step includes:
[0020] After the number of frames processed in the single-frame judgment step reaches a preset number, the number of frames within the preset number of frames is counted to determine whether the sensor is installed in a front-facing, side-mounted, or top-mounted configuration; and
[0021] The sensor mounting method is determined by comparing the number of frames mounted on the front, side, or top with a preset range of frames.
[0022] Optionally, the sensor has a boundary detection module adapted to various installation methods, "performing boundary detection of the area to be detected based on the sensor's installation method," and the boundary detection methods include:
[0023] The boundary detection module, adapted to the installation method, performs boundary detection on the area to be detected.
[0024] Optionally, "testing the area to be tested" includes:
[0025] Acquire at least two frames of raw target data;
[0026] Boundary detection processing is performed on each frame of the original target data to obtain the boundary detection results for each frame of the region to be detected; and
[0027] Multi-frame comprehensive quantization detection is performed on the boundary detection results of a preset number of frames to determine the target boundary detection results of the area to be detected.
[0028] To achieve the aforementioned objectives, this application also provides a boundary detection device for boundary detection using sensors positioned within the detection area, comprising:
[0029] The acquisition module is used to acquire at least two frames of raw target data;
[0030] The installation method determination module is used to analyze at least two frames of raw target data to determine the sensor's installation method; and
[0031] The boundary detection module can perform boundary detection on the area to be detected based on the sensor installation method;
[0032] Optionally, the sensor can be installed in a front-mounted, side-mounted, or top-mounted manner.
[0033] Optionally, the sensor is a millimeter-wave radar.
[0034] To achieve the aforementioned objectives, this application also provides a target detection method for detecting targets within a specific region having physical boundaries, comprising:
[0035] The boundary detection method or device described above is used to obtain the boundary detection results of a specific area; and
[0036] Target detection is performed based on boundary detection results.
[0037] To achieve the above-mentioned objectives, this application also provides a boundary detection device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the boundary detection method and / or target detection method described above.
[0038] To achieve the above-mentioned objectives, this application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the boundary detection method and / or target detection method described above.
[0039] To achieve the above-mentioned objectives, this application also provides an apparatus, comprising:
[0040] The equipment body; and the detection device as described above, which is installed on the equipment body;
[0041] The sensors are used for boundary / target detection and / or for communication.
[0042] The aforementioned boundary detection method, apparatus, storage medium, and device allow the server to acquire at least two frames of original target data in the area to be detected, determine the sensor installation method based on the analysis of the at least two frames of original target data, and perform boundary detection on the area to be detected based on the sensor installation method. Thus, the sensor installation position can be determined and the boundary of the area to be detected can be automatically detected without human intervention. The method is simple, has high detection efficiency, and saves detection costs. Attached Figure Description
[0043] Figure 1 This is a flowchart illustrating a boundary detection method in one embodiment;
[0044] Figure 2 This is a schematic diagram of the radar system in one embodiment;
[0045] Figure 3 This is a flowchart illustrating a boundary detection method in one embodiment;
[0046] Figure 4 This is a flowchart illustrating a boundary detection method in one embodiment;
[0047] Figure 5 This is a flowchart illustrating a boundary detection method in one embodiment;
[0048] Figure 6 This is a flowchart illustrating a boundary detection method in one embodiment;
[0049] Figure 7 This is a flowchart illustrating a target detection method in one embodiment;
[0050] Figure 8 This is a structural block diagram of a boundary detection device in one embodiment;
[0051] Figure 9 This is a structural block diagram of a boundary detection device in one embodiment;
[0052] Figure 10 This is a structural block diagram of a boundary detection device in one embodiment;
[0053] Figure 11 This is a structural block diagram of a boundary detection device in one embodiment;
[0054] Figure 12 This is a structural block diagram of a boundary detection device in one embodiment;
[0055] Figure 13 This is an internal structural diagram of a boundary detection device in one embodiment. Detailed Implementation
[0056] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0057] When using sensors to detect targets in specific areas (such as rooms, parking lots (indoor or outdoor)), the actual field of view boundary of the sensor is generally determined by the sensor's field of view and the physical boundary of the specific area after the sensor is fixedly installed. Therefore, before the sensor can work normally, the actual field of view boundary must be determined to reduce or even eliminate false targets caused by multipath effects, so as to achieve more accurate boundary detection or tracking in the actual field of view area.
[0058] For example, when a sensor is used indoors for boundary detection and tracking, it needs to be determined based on the sensor's own field of view and the physical boundary of the area covered by the field of view. Therefore, it is necessary to first obtain the actual field of view boundary, such as the boundary, area, corners, and shape of the room in the sensor's field of view.
[0059] In practical applications, users have certain personalized requirements for the installation method of sensors or devices equipped with sensors. For example: front installation (hereinafter referred to as front mounting), such as installation on the front of a wall-mounted air conditioner; side installation (hereinafter referred to as side mounting), such as installation on the side of a floor-standing air conditioner; top installation (hereinafter referred to as top mounting), such as installation on the top of a ceiling-mounted air conditioner.
[0060] Indoor boundary detection requires prior knowledge of the sensor's installation location, and then the boundary detection parameters and / or strategies should be adjusted accordingly based on the different sensor installation locations. In the existing technology, obtaining the sensor's installation location requires the participation of installation personnel.
[0061] Example 1
[0062] After understanding the aforementioned technical problems, the inventors of this application have creatively proposed a boundary detection method through research and analysis. This method can be used for boundary detection of sensors placed within a detection area, determining the sensor's installation position without human intervention. The detection area can have at least three physical boundaries (hereinafter referred to as boundaries). In this embodiment, the physical boundaries can be comprehensively defined based on any combination of the sensor's own field of view, the field of view coverage area, the sensor's own physical boundary definition logic, and the operating mode. For example, different sensors may define the physical boundaries of the same detection area differently, and the same sensor may also define the physical boundaries of the same detection area differently depending on the specific application scenario. Figure 1 As shown, the boundary detection method may include the following steps:
[0063] S1. Acquire at least two frames of raw target data. The raw target data described in this application may also be referred to as raw target points. For example, the raw target data may be at least two frames of echo signals reflected by the target, or data after certain signal processing of the echo signals.
[0064] The original target data can be the data corresponding to the region to be detected. The region to be detected can be any polygonal region, such as a triangle, quadrilateral, pentagon, hexagon, or irregular shape. When the region to be detected is a triangle, its three sides can be considered as three physical boundaries. When the region to be detected is an irregular shape, one implementation is to treat it as a polygon whose outline is similar to that of the irregular shape.
[0065] In this embodiment, the sensor can be placed at any location in the area to be detected. For example, the sensor can be placed in a corner of the area to be detected, i.e., the lateral installation mentioned above. The server acquires the original target data of the area to be detected collected by the sensor. The sensor can acquire the data by transmitting a composite beam in the air through multiple transmitting antennas in the elevation dimension. For example, three transmitting antennas can be used to perform spatial beamforming to form an electromagnetic wave signal with a narrow elevation beam. The sensor receives the echo signals reflected from surrounding targets. Multiple (e.g., four) receiving antennas in the azimuth dimension of the sensor receive the echo signals and perform signal processing on the echo signals, such as range dimension, velocity dimension, incoherent accumulation, constant false alarm rate detection, angle measurement (azimuth), etc., to obtain multiple frames of original target data. Optionally, each frame of original target data includes information such as the target's range, velocity, and azimuth. This embodiment does not limit this.
[0066] The signal processing of the echo signal described above may include various analog signal processing operations (such as demodulation, filtering, mixing, sampling, AD conversion, etc.) and digital signal processing operations (such as Fourier transform (FFT), constant false alarm rate detection (CFAR), direction of arrival (DOA), etc.) to obtain raw target data containing information such as target velocity, distance, and direction angle (such as azimuth and elevation angle). The raw target data may represent various target types, such as moving targets, stationary targets, and false targets caused by multipath effects.
[0067] S2. Based on the processing of at least two frames of raw target data, determine the sensor installation method.
[0068] Optionally, the analysis of at least two frames of original target data can be performed by denoising each frame of original target data to obtain effective target data.
[0069] Since the original target data contains noise data that affects the determination of the sensor's installation method, noise such as those within the sensor's blind zone, outside the environmental limitation range, and / or in motion can be removed based on the requirements for determining the sensor's installation method, environmental characteristics, and the sensor's own performance limitations. For example, when the sensor is used indoors, target points located within the sensor's blind zone, in motion, or close in front can be considered as noise, and the range of such noise can be set according to actual needs.
[0070] It should be noted that environmental constraints can be understood as prior knowledge of the sensor's application scenario. For example, if the application scenario is in a typical resident's bedroom, the distance in the original target data can be limited to the length of a diagonal line across the bedroom, such as 10 meters. Targets with a distance greater than 10 meters are considered false targets caused by multipath propagation and are removed. In other words, based on pre-defined information, certain constraints can be set before boundary detection to further improve its efficiency and accuracy.
[0071] Meanwhile, the aforementioned "being in motion" can be understood as a target point whose velocity is not zero, or whose velocity is greater than a preset value (such as the sensor boundary detection velocity error value, or other preset values). A near-field target point can be understood as a boundary target that would not normally exist within a very close range of the sensor; for example, when the sensor is installed in the middle of the ceiling of a living room, the distance between the living room boundary and the sensor is generally greater than 1 meter. Therefore, near-field target points less than 1 meter from the sensor can be considered as noise and filtered out.
[0072] In this embodiment, the method for determining the sensor installation method based on valid target data is not limited. It can be the method shown in other embodiments of this application, the method that can be learned by those skilled in the art under the guidance of this application, or the method that those skilled in the art can adapt to implement the technical solution of this application.
[0073] S3. Based on the sensor installation method, perform boundary detection on the area to be detected.
[0074] Since the boundary detection modules used for sensors differ depending on their installation method, the appropriate boundary detection module can be adapted to perform boundary detection on the area to be detected once the sensor installation method is known.
[0075] Optionally, the adaptation process can be automatically executed by the sensor's control unit in response to the sensor's mounting method, thereby achieving adaptive border inspection.
[0076] In this embodiment, the method of boundary detection is not limited. It can be the method shown in other embodiments of this application, the method that can be learned by those skilled in the art under the guidance of this application, or the method that those skilled in the art can adapt to implement the technical solution of this application.
[0077] Optionally, the sensor mentioned in any of the embodiments below in this embodiment can further be a radar, more specifically a millimeter-wave radar, for example, a 77GHz millimeter-wave radar with a 4GHz bandwidth design, or it can also be used for 60GHz, etc.
[0078] This embodiment also provides a radar system, such as Figure 2 As shown, it includes the radar 102 described above and a server 101 that is connected to the radar for communication.
[0079] Server 101 can be a standalone server or a server cluster consisting of multiple servers. Radar 102 can be a millimeter-wave radar, for example, a 77GHz millimeter-wave radar with a 4GHz bandwidth design. It should be noted that radar 102 has transceiver antennas for measuring angle and distance. Optionally, the radar also has an antenna for transmitting beamforming electromagnetic wave signals to form a narrow elevation beam.
[0080] Optionally, server 101 and radar 102 can be installed on the same device, for example, both can be integrated into a smart device (such as air conditioner, lamp, electric fan, robot vacuum cleaner, electric heater, etc.), or even integrated on the same PCB board as a single device for easy installation on electronic devices to enable intelligent operation of electronic devices.
[0081] Optionally, the server 101 described above can be integrated with the radar 102 or even into a SoC chip, such as an AiP (Antenna-in-Package) chip or an AoC (Antenna-on-Chip) chip.
[0082] It should be noted that the execution subject of the boundary detection method provided in this embodiment can be a server 101 or a boundary detection device capable of executing the above-described boundary detection method. The boundary detection device can be part or all of the server 101 or radar 102 through software, hardware, or a combination of software and hardware. In the following method embodiments, the execution subject is described using the server 101 as an example.
[0083] In summary, the boundary detection method provided in this embodiment can determine the installation position of the sensor and automatically detect the boundary of the area to be detected without human intervention. The method is simple, efficient, and saves detection costs.
[0084] Example 2
[0085] This embodiment provides a specific implementation example of the boundary detection method mentioned in Embodiment 1. Specifically, refer to... Figure 3 As mentioned in S2 of Example 1: The method of obtaining the sensor installation based on the analysis of at least two frames of raw target data includes the following steps:
[0086] S21. Single-frame judgment step: Analyze a single frame of original target data to determine the sensor installation method.
[0087] The analysis based on a frame of raw target data can be used to distinguish the sensor installation method based on the angle range of each boundary relative to the sensor (because the angle range of each boundary relative to the sensor is different for different installation methods, and this angle range can be obtained through experiments or experience. In some embodiments of this application, the angle range is obtained based on empirical values as an example). It can also be other methods that can determine the sensor installation method. This embodiment does not limit this method.
[0088] Optionally, when the analysis based on at least one frame of original target data is instead based on two or more frames of original data, the method further includes:
[0089] S22, Multi-frame judgment step: Based on the judgment results of the original target data of the preset number of frames in S21, quantitative analysis is performed to finally confirm the sensor installation method.
[0090] The preset frame count refers to the number of frames set in advance. This preset frame count is determined based on the actual situation, and its purpose is to increase the sample size of the sensor installation method judgment results and improve the accuracy of the judgment results. The multi-frame judgment step refers to the quantitative analysis of the sensor installation method judgment results obtained from S21 based on all frames of the preset frame count, and based on the analysis results, the sensor installation method is finally confirmed.
[0091] The boundary detection method provided in this embodiment confirms the sensor installation method by quantitatively analyzing the judgment results of a single frame of original target data. This expands the sample size of the sensor installation method judgment results and improves the accuracy of the judgment results.
[0092] Example 3
[0093] This embodiment illustrates a specific implementation example of the single-frame judgment step shown in Embodiment 2, referencing... Figure 4 The steps include:
[0094] S211. On the plane of the coordinate system where the sensor is located, obtain the initial boundary model of multiple boundaries of the region to be detected based on the original target data of the current frame. For example, obtain at least three initial boundary models of at least three boundaries of the region to be detected based on the original target data of the current frame.
[0095] The initial boundary model can be a linear model corresponding to each boundary constructed according to the actual situation. Taking the boundary of the area to be detected as a straight line as an example, the server can obtain the straight line model of the boundary of the area to be detected based on the plane (XY plane) of the Cartesian coordinate system where the sensor is located.
[0096] In this embodiment, optionally, the straight line model corresponding to each boundary of the region to be detected obtained by the server can be:
[0097] Ax + By + C = 0
[0098] Where A = cosθ, B = sinθ, C = d, and θ is the straight line containing the boundary to be detected corresponding to this linear model. For example, it could be... Figure 2 The angle between the straight line containing boundary 1 and the vertical axis in the rectangular coordinate system, where the direction of the vertical axis is the normal direction of the sensor, could be, for example, the angle between the straight line containing boundary 1 and the vertical axis in the rectangular coordinate system. Figure 2 The angle between the radar 102 and the boundary 1 is approximately 45°, and it points in the direction of the area to be detected; d is the distance between the straight line corresponding to this straight line model and the origin in the rectangular coordinate system.
[0099] S212. For the initial boundary model of each boundary, perform multiple boundary searches within a preset angle range and a preset distance range to determine the boundary parameters of each boundary in the area to be detected.
[0100] The preset angle range refers to the angle range corresponding to each boundary, where the angle is the angle between each boundary and the vertical axis in the plane coordinate system. The preset distance range refers to the distance range corresponding to each boundary, where the distance is the distance between each boundary and the origin in the plane coordinate system. The confidence level refers to the ratio of the number of original target points whose distance from the boundary to the boundary is less than a preset distance threshold to the total number of targets. The angle range and distance range are determined based on the actual situation of the area to be detected. The boundary parameters include the confidence level. Optionally, the boundary parameters may also include angle and / or distance, where the angle may fall within the aforementioned preset angle range, and the distance may fall within the aforementioned preset distance range.
[0101] In this embodiment, for the i-th boundary, optionally, the server searches for values within its corresponding angular range using different angle values and a certain distance value to obtain the boundary parameters of the i-th boundary corresponding to different angles; similarly, the server searches for values within its corresponding distance range using different distance values and a certain angle value to obtain the boundary parameters of the i-th boundary corresponding to different distances. Preferably, when the server searches by distance and angle respectively, it can set a distance step size and an angle step size, and use the distance step size and angle step size as a reference to perform the boundary search to obtain the boundary parameters corresponding to multiple boundary models of the i-th boundary. It is understood that the server can be a local server or a cloud server with data processing capabilities.
[0102] Optionally, the server can determine the boundary parameter of the boundary model with the highest confidence level as the boundary parameter of the i-th boundary based on the confidence levels of multiple boundary models of the i-th boundary; the server can also determine the boundary parameter of the boundary model with the smallest product of angle and distance as the boundary parameter of the i-th boundary based on the relationship between angle and distance, such as the product of angle and distance. This embodiment does not limit this.
[0103] Specifically, taking three boundaries—boundary 1, boundary 2, and boundary 3—as an example, the process of determining the boundary parameters for each boundary is as follows:
[0104] (a) Based on the linear model corresponding to each boundary described above, θ searches M times with a step size of Δθ within the range [θ1, θ2], and d searches N times with a step size of Δd within the range [d1, d2]. Therefore, the boundary search model searches within the two-dimensional plane (θ, d), such that the m-th search angle of θ is θ. m =m·Δθ, d The vertical distance of the nth search is d n =n·Δd, at this time the boundary line model is L mn A m +B m +C n =0,
[0105] Among them, A m =cos(θ) m B m =sin(θ) m C n =d n
[0106] (b) Calculate the detection points P in the XY plane k :(x k ,y k ) to L mn Distance D mnk Statistics D mnk <Tr Number of detection points J mn ;
[0107] (c) Repeat step (b) to search for max(J). mn The maximum value J in m∈[0,M],n∈[0,N] 1max This determines the parameters L1:θ of boundary 1. 1max d 1max and confidence level J 1max ;
[0108] (d) Repeat steps (a), (b), and (c) to calculate the parameters of boundary 2 and boundary 3 respectively. The search ranges of θ and d for each boundary are different: [θ1,θ2] and [d1,d2], depending on the boundary type.
[0109] S213, Select the J with the highest confidence level 1max The boundary is used as the feature boundary, and the installation method of the sensor is confirmed based on the correspondence between the angle of the feature boundary and the preset angle range.
[0110] The boundary types mentioned above are boundary 1, boundary 2, and boundary 3. In some implementations, each boundary can be redefined or renamed as needed to distinguish between them.
[0111] Taking three boundaries as an example, namely boundary 1, boundary 2, and boundary 3, in this embodiment, the boundary with the highest confidence is taken as the feature boundary. If the feature boundary is boundary 1, its boundary information is: θ feature d feature J feature .
[0112] In this embodiment, multiple angle ranges are set for various installation methods of the sensor. The installation method of the sensor is determined by the correspondence between the angle information of the feature boundary and the multiple angle ranges.
[0113] The sensor installation method is determined based on feature boundary information, and the specific method is as follows:
[0114] (a) If θ feature ∈[θ f1 ,θ f2 If the sensor is installed in the correct position, it is determined that the sensor is installed in the front (or upright).
[0115] (b) If θ feature ∈[θ s1 ,θ s2 If the sensor is installed on the side, it is determined that the sensor is installed on the side (or side-mounted).
[0116] (c) If θ feature∈[θ t1 ,θ t2 If the sensor is installed in the top position (or top mount), then the sensor is determined to be installed on the top.
[0117] Where, θ f1 θ f2 θ s1 θ s2 θ t1 θ t2 It is determined based on experience or is a preset value.
[0118] It is understood that the front mounting described in the embodiments of this application refers to the angle θ between the normal direction of the sensor and the boundary 1. feature Approximately 90° (or any angle within the range of 90° ± 15°, if equal to 90°). Side mounting refers to the angle θ between the sensor's normal direction and boundary 1. feature Approximately 45° (or any angle within the range of 45° ± 15°, if equal to 45°). For the included angle θ... feature Within the range of [0°, 90°], except for [30°, 60°] and the other angle ranges (0°, 30°) and (60°, 75°) except for [75°, 90°], it is determined to be top-mounted.
[0119] The sensor installation information obtained from each frame of raw target data will be saved for later retrieval.
[0120] In this embodiment, multiple initial boundary models are established for each boundary, and the boundary model with the highest confidence is selected as the feature boundary and its boundary parameters are used as a reference. This setting can improve the accuracy of the boundary parameters of each boundary, thereby improving the accuracy of the sensor installation method determination.
[0121] Example 4
[0122] This embodiment illustrates a boundary detection method, providing a specific implementation example of S22 (multi-frame judgment step) in Embodiment 2. For ease of understanding, please refer to... Figure 5 In conjunction with Embodiment 1 and Embodiment 2, the method includes the following steps:
[0123] S221. In response to the number of frames processed in the single frame judgment step reaching the preset number of frames, count the number of frames within the preset number of frames that determine the sensor installation method as front mounting, side mounting, and top mounting.
[0124] In response to the number of frames processed in S21 (single frame determination step) reaching a preset number of frames, for example, M FIf a frame is not found, the process proceeds to the multi-frame judgment step, which is as follows: The number of frames determined to be front-mounted, side-mounted, or top-mounted from the preset frame count is statistically analyzed.
[0125] (a)M F Within a frame, the number of frames that are statistically determined to be in normal condition is M. Jf ;
[0126] (b)M F Within a frame, the statistical decision is based on the number of frames M that are side-mounted. Js ;
[0127] (c)M F Within a frame, the statistical decision is the number of frames M that are top-loaded. Jt .
[0128] S222. Determine the sensor mounting method based on the comparison results of the number of frames for front mounting, side mounting, and top mounting with the preset number of frames range.
[0129] The preset frame range can be determined based on actual conditions, with the aim of improving the accuracy of the judgment results through quantitative analysis. The specific implementation method is as follows:
[0130] (a) If M Jf ∈[M Jfmin M Jfmax The final ruling was that it was formal attire;
[0131] (b) If M Js ∈[M Jsmin M Jsmax The final ruling was side-mounted;
[0132] (c) If M Jt ∈[M Jtmin M Jtmax The final judgment was that it should be installed at the top.
[0133] Among them, M Jfmin M Jfmax M Jsmin M Jsmax M Jtmin M Jtmax It is obtained from experience or is a preset value.
[0134] For example, it can be targeted at M F The original target data of the frame is processed, where M F ≥2, e.g., M FWhen the value equals 100, a single-frame analysis is performed on each of the 100 frames (see the steps above), resulting in 100 corresponding installation method determination results. This means the sum of the statistical counts for the three installation methods is 100. Multi-frame judgments are then performed based on these 100 results. Please refer to steps S221 and S222 above for details. For example, if the number of frames with a single-frame judgment of "correct installation" is greater than 50 out of the 100 results (i.e., the number of frames with a single-frame judgment of "correct installation" > M), then... F If half of the frames are positive, the final result of the radar installation method determination is considered to be upright installation. Similarly, for multi-frame determination of side-mounted installation, please refer to the multi-frame determination of upright installation. If the multi-frame determination result is neither upright nor side-mounted, then the multi-frame determination result is top-mounted. For example, if after 100 frames of multi-frame determination, 52 frames are upright installation, 30 frames are side-mounted, and 18 frames are top-mounted, then the multi-frame determination result is upright installation.
[0135] In this embodiment, the sensor installation method determination information obtained in the single-frame determination step can be used to make a final determination of the sensor installation method through quantitative analysis, thereby further improving the accuracy of the sensor installation method determination.
[0136] Example 5
[0137] This embodiment provides a boundary detection method, which gives a specific implementation example of boundary detection of the region to be detected as described in S3 of Embodiment 1. For ease of understanding, please refer to... Figure 2 , Figure 6 In conjunction with Examples 1 to 4, the method includes:
[0138] S31. Obtain at least two frames of raw target data.
[0139] In this application, the at least two frames of original target data correspond to the target data of the region to be detected. Furthermore, the at least two frames of original target data are temporally consecutive frames.
[0140] In this embodiment, taking radar as a specific sensor as an example, the server acquires the original target data of the area to be detected collected by the radar. The radar acquires the data by transmitting a composite beam in the air through multiple (e.g., 3) transmitting antennas in the elevation dimension to form a narrow-beam electromagnetic wave signal in elevation. It receives the echo signals reflected from surrounding targets. In the azimuth dimension, multiple (e.g., 4) receiving antennas in the radar receive the echo signals and perform signal processing on the echo signals, such as range dimension, velocity dimension, incoherent accumulation, constant false alarm rate detection, angle measurement (azimuth), etc., to obtain multiple original target point data. Optionally, each original target data includes information such as the target's range, velocity, and azimuth, but this embodiment does not limit this.
[0141] S32. Perform boundary detection processing on each frame of the original target data to obtain the boundary detection results of each frame of the region to be detected.
[0142] The boundary detection results include the boundary information and boundary status of each boundary in the region to be detected. The boundary information refers to the area of the region to be detected formed by the boundary, the boundary model corresponding to each boundary, and the coordinates of the inflection points of the region to be detected formed by each boundary, etc. The boundary status refers to whether the current boundary is valid.
[0143] In this embodiment, since the radar acquires multiple frames of raw target data, the server can perform single-frame boundary detection operations, that is, perform boundary detection operations on each boundary of the region to be detected based on the raw target data of each frame. Optionally, the server can construct an initial boundary model for each boundary of the region to be detected corresponding to the current frame based on the raw target data of each frame. Based on the raw target data, the initial boundary model of each boundary corresponding to the current frame is optimized by linear fitting to obtain the boundary parameters corresponding to each boundary of the current frame. The server can also search and construct multiple candidate boundary models based on the initial boundary model corresponding to the current frame, with a preset angle step size and a preset distance step size. Based on the distance relationship between each candidate boundary model and the raw target data, the candidate boundary model corresponding to each boundary of the current frame is finally determined. Based on the candidate boundary model, the boundary parameters corresponding to each boundary are determined. This embodiment does not limit this.
[0144] S33. Perform multi-frame comprehensive quantization detection on the boundary detection results of the preset number of frames to determine the target boundary detection results of the area to be detected.
[0145] The preset frame count refers to the number of frames set in advance. This preset frame count is determined based on the actual situation, and its purpose is to increase the sample size for boundary detection and improve the accuracy of the boundary detection results. Multi-frame comprehensive quantization detection refers to further quantifying the effectiveness of the boundary parameters of each boundary corresponding to all frames within the preset frame count.
[0146] In this embodiment, to improve the efficiency of server-side comprehensive quantization, during the process of valid quantization of boundary parameters for all boundaries corresponding to a preset number of frames, the server can first perform validity filtering based on the boundary detection results of each frame, retaining only valid single-frame boundary detection results, and then perform comprehensive quantization based on the valid single-frame boundary detection results. Specifically, comprehensive quantization can be performed on each boundary from aspects such as the angle between each boundary and the vertical axis of the plane, the distance between each boundary and the origin, the angle between adjacent boundaries, and the distance between opposite edges. Ultimately, the target boundary detection result for the current detection area is determined. This target boundary detection result can include the boundary information and boundary state of each boundary in the detection area; this embodiment does not limit this aspect.
[0147] In the aforementioned boundary detection method, the server acquires at least two frames of raw target data, performs boundary detection processing on the raw target points in each frame, and obtains the boundary information and boundary state of the target area for each frame, which serves as the boundary detection result for the current frame. Then, multi-frame comprehensive quantization detection is performed on the boundary detection results of a preset number of frames to determine the target boundary detection result for the target area. In this method, the server integrates a single-frame detection algorithm module and a multi-frame comprehensive quantization algorithm module for boundary detection. Based on the raw target points of the target area automatically scanned and collected by radar, boundary detection processing is performed. This eliminates the need to acquire human movement data to obtain corresponding indoor boundary detection data, simplifying the boundary detection process and improving its efficiency.
[0148] Example 6
[0149] This embodiment provides a target detection method, referencing... Figure 7 The method includes the following steps:
[0150] S01. Obtain the boundary detection result by applying the boundary detection method provided in any of the embodiments from Embodiment 1 to Embodiment 5, or any combination of embodiments that can be combined.
[0151] When using sensors to detect targets in specific areas (such as rooms, parking lots (indoor or outdoor)), the actual field of view boundary of the sensor is generally determined by the sensor's field of view and the physical boundary of the specific area after the sensor is fixedly installed. Therefore, before the sensor can work normally, the actual field of view boundary must be determined to reduce or even eliminate false targets caused by multipath effects, so as to achieve accurate target detection or tracking in the actual field of view area.
[0152] In this embodiment, the method for obtaining boundary detection results can determine the sensor installation method without human intervention, which facilitates the acquisition of boundary detection results, thereby improving the efficiency of target detection and reducing the cost of target detection.
[0153] S02. Target detection is performed based on the boundary detection results obtained in S01.
[0154] The target detection mentioned here is based on the boundary detection results, which can improve the accuracy of target detection. In this embodiment, the specific implementation method of target detection is not limited.
[0155] Example 7
[0156] This embodiment provides a boundary detection device 10, used for boundary detection by sensors placed within the detection area. (Refer to...) Figure 8 It includes an acquisition module 11, an installation method determination module 12, and a boundary detection module 13.
[0157] The acquisition module 11 is used to acquire at least two frames of raw target data; the installation method determination module 12 is used to analyze at least two frames of raw target data to determine the installation method of the sensor; the boundary detection module 13 is able to perform boundary detection on the area to be detected based on the installation method of the sensor; wherein, the installation method of the sensor includes at least two of the following: upright mounting, side mounting, and top mounting.
[0158] In one embodiment, such as Figure 9 As shown, the boundary detection module 13 includes
[0159] Creation unit 131, search unit 132, and determination unit 133, wherein:
[0160] Create unit 131 to obtain the initial boundary model of each boundary of the region to be detected based on the original target data of the current frame on the plane of the coordinate system where the sensor is located;
[0161] Search unit 132 is used to perform multiple boundary searches within a preset angle range and a preset distance range for the initial boundary model of each boundary, and to determine the boundary parameters of each boundary of the region to be detected.
[0162] The determination unit 133 is used to determine the effective boundary model of the region to be detected based on the boundary parameters of each boundary and the relationship between adjacent boundaries;
[0163] The search unit 132 is also used to perform boundary search based on the angle of the effective boundary model to obtain the boundary detection result corresponding to the original target data of the current frame;
[0164] The boundary detection results include boundary information and / or boundary status for each boundary in the area to be detected; and / or, the boundary parameters include confidence level, and optionally, the boundary parameters may also include angle and / or distance.
[0165] In one embodiment, the search unit 132 is specifically configured to perform M boundary searches within a preset angle range with a preset angle step size, and N boundary searches within a preset distance range with a preset distance step size, for each boundary, based on an initial boundary model, to obtain M*N candidate boundary models corresponding to each boundary; calculate the distance from the original target point represented by each original target data in the current frame to each candidate boundary model of each boundary; and determine the boundary parameters of each boundary based on the distance.
[0166] Where M and N are positive integers.
[0167] In one embodiment, the search unit 132 is specifically used to count the number of original target points represented by the original target data whose distance is less than a preset distance threshold for the candidate boundary model of the i-th boundary; and determine the boundary parameters of the i-th boundary based on the candidate boundary model with the largest number of original target points.
[0168] Where i is a positive integer.
[0169] In one embodiment, the search unit 132 is specifically configured to, if a valid boundary model exists in the region to be detected, search for adjacent boundary models of the valid boundary model from the perspective of the valid boundary model, and search for the remaining boundary models from the perspectives of the valid boundary model and the adjacent boundary models, output the boundary information corresponding to the valid boundary model, the adjacent boundary models and the remaining boundary models, and determine that the boundary state is valid; if no valid boundary model exists, output the boundary information corresponding to the default boundary model, and determine that the boundary state is invalid.
[0170] In one embodiment, when the region to be detected has an even number of boundary models, the search unit 132 is specifically configured to: if there is a valid boundary model, search for the adjacent boundary models of the valid boundary model from the perspective of the valid boundary model, and search for the remaining boundary models from the perspective of the valid boundary model and the adjacent boundary models, output the boundary information corresponding to the valid boundary model, the adjacent boundary models, and the remaining boundary models, and determine that the boundary state is valid; if there is an odd number of pairwise adjacent valid boundary models, search for the remaining boundary models from the perspective of the valid boundary models, output the boundary information corresponding to the valid boundary model and the remaining boundary models, and determine that the boundary state is valid; if there is no valid boundary model, determine that the boundary state is invalid.
[0171] In one embodiment, such as Figure 10As shown, the boundary detection device 10 also includes an output module 14;
[0172] Output module 14 is used to determine the validity of boundary information based on boundary information, adjacent boundary relationships and distances to opposite edges if the boundary state is valid. If the boundary information is valid, the boundary information is output and the boundary state is determined to be valid. If the boundary information is invalid, the boundary information corresponding to the default boundary model is output and the boundary state is determined to be invalid. If the boundary state is invalid, the boundary information corresponding to the default boundary model is output and the boundary state is determined to be invalid.
[0173] In one embodiment, such as Figure 11 As shown, the boundary detection device 10 further includes a comprehensive quantization module 15, comprising a selection unit 151, a judgment unit 152, and an output unit 153, wherein:
[0174] Selection unit 151 is used to obtain the number of frames with valid boundary states. If the number of frames is greater than a preset number threshold, then for each boundary, the boundary information with the highest confidence is selected from the boundary information of each frame as the target boundary information. The number threshold is determined based on the decision factor and the preset number of frames.
[0175] The judgment unit 152 is used to judge the validity of the target boundary information based on the target boundary information, the relationship between adjacent boundaries and the distance between opposite edges. If the target boundary information is valid, the target boundary information is output and the boundary state is determined to be valid.
[0176] Output unit 153 is used to output the target boundary information corresponding to the default boundary model if the target boundary information is invalid, thus determining that the boundary state is invalid.
[0177] In one embodiment, such as Figure 12 As shown, the boundary detection device 10 also includes a data processing module 16;
[0178] The data processing module 16 is used to preprocess the original target points and remove noise from the original target points; the noise includes at least one of the following: noise in the radar blind zone of the area to be detected, moving points, left and right noise, and noise in the preset area.
[0179] Specific limitations regarding the boundary detection device can be found in the limitations of the boundary detection method described above, and will not be repeated here. Each module in the aforementioned boundary detection device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in hardware or independently of the processor in the boundary detection device, or stored in software in the memory of the boundary detection device, so that the processor can call and execute the operations corresponding to each module.
[0180] In one embodiment, a boundary detection device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 13 As shown, the boundary detection device includes a processor, memory, communication interface, display screen, and input device connected via a system bus. The processor provides computational and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a boundary detection method. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the device's casing, or an external keyboard, touchpad, or mouse.
[0181] Example 8
[0182] When using sensors to detect targets in specific areas (such as rooms, parking lots (indoor or outdoor)), the actual field of view boundary of the sensor is generally determined by the sensor's field of view and the physical boundary of the specific area after the sensor is fixedly installed. Therefore, before the sensor can work normally, the actual field of view boundary must be determined to reduce or even eliminate false targets caused by multipath effects, so as to achieve precise target detection or tracking in the actual field of view area.
[0183] For example, when a sensor is used indoors for target detection and tracking, it needs to be determined based on the sensor's own field of view and the physical boundary of the area covered by the field of view. Therefore, it is necessary to first obtain the actual field of view boundary, such as the boundary, area, corners, and shape of the room in the sensor's field of view.
[0184] In one embodiment, a boundary detection device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to perform the following steps:
[0185] Acquire at least two frames of raw target points;
[0186] Boundary detection processing is performed on the original target points of each frame to obtain the boundary detection results of each frame of the region to be detected; the boundary detection results include the boundary information and boundary state of each boundary of the region to be detected.
[0187] Multi-frame comprehensive quantization detection is performed on the boundary detection results of a preset number of frames to determine the target boundary detection results of the area to be detected.
[0188] The boundary detection device provided in the above embodiments has a similar implementation principle and technical effect to the above method embodiments, and will not be described again here.
[0189] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:
[0190] Acquire at least two frames of raw target points;
[0191] Boundary detection processing is performed on the original target points of each frame to obtain the boundary detection results of each frame of the region to be detected; the boundary detection results include the boundary information and boundary state of each boundary of the region to be detected.
[0192] Multi-frame comprehensive quantization detection is performed on the boundary detection results of a preset number of frames to determine the target boundary detection results of the area to be detected.
[0193] The computer-readable storage medium provided in the above embodiments has similar implementation principles and technical effects to the above method embodiments, and will not be described again here.
[0194] In another optional embodiment, the detection device described in this application can be a sensor or part of a sensor (such as radar equipment or integrated circuits, such as millimeter-wave radar chips). When the detection device is a sensor, it can be fixedly installed within the detection area to facilitate target detection. The sensor may include a transmitting and receiving antenna, a control unit, and a processing unit. The transmitting and receiving antenna is used to transmit and receive radio signals to facilitate target detection; the processing unit is used to process the echo signals to obtain target information. The aforementioned transmitting and receiving antenna may include at least three transmitting antennas, and these at least three transmitting antennas are located on non-collinear lines (i.e., not collinear) to facilitate the detection of elevation and azimuth angles.
[0195] Optionally, the aforementioned sensor may further include a control unit for controlling the at least three transmitting antennas to transmit radio signals. When the sensor is fixed within the detection area, the control unit controls the transmitting antennas to emit a test signal for boundary detection, which may be the same as or different from the signal subsequently used for target detection. Simultaneously, the receiving antenna in the transmitting and receiving antennas receives the test echo signal for boundary detection, and the processing unit performs operations such as signal processing and data processing on the test echo signal to obtain boundary information on a preset two-dimensional plane.
[0196] In another optional embodiment, in order to further improve the detection accuracy of boundary information on a preset two-dimensional plane, the control unit can also control the at least three transmitting antennas to transmit narrow-beam test signals in a preset dimension, that is, the preset dimension is perpendicular to the two-dimensional plane.
[0197] In an optional embodiment, the processing unit described above may also use the method described in any embodiment of this application to process the test echo signal to obtain the boundary information on the preset two-dimensional plane. For details, please refer to the relevant sections of this application, which will not be repeated here.
[0198] In an optional embodiment, this application also provides a device, which may include: a device body; and a sensor as described in any embodiment of this application disposed on the device body; wherein the sensor is used for target detection and / or communication.
[0199] Optionally, the aforementioned equipment can be intelligent transportation equipment (such as vehicles, bicycles, motorcycles, ships, subways, trains, etc.), security equipment (such as cameras), liquid level / flow rate detection equipment, smart wearable devices (such as wristbands, glasses, etc.), smart home devices (such as televisions, air conditioners, smart lights, etc.), various communication devices (such as mobile phones, tablets, etc.), as well as devices such as road gates, intelligent traffic lights, intelligent signs, traffic cameras, and various industrial robotic arms (or robots), and can also be various instruments used to detect vital signs parameters and various devices equipped with such instruments.
[0200] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.
[0201] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above 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.
[0202] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.
Claims
1. A boundary detection method, characterized in that, The method for boundary detection using sensors positioned within the detection area includes: Acquire at least two frames of raw target data; Based on the analysis of the at least two frames of raw target data, the installation method of the sensor is determined; and Based on the installation method of the sensor, boundary detection is performed on the area to be detected; The sensor can be installed from the top, side, or front.
2. The method according to claim 1, characterized in that, The step of obtaining the sensor's installation method based on the analysis of at least two frames of raw target data includes: The single-frame judgment step analyzes a single frame of raw target data to determine the installation method of the sensor. as well as The multi-frame judgment step performs quantitative analysis based on the judgment results of the installation method in the preset number of frames in the single-frame judgment step to finally confirm the installation method of the sensor.
3. The method according to claim 2, characterized in that, The single-frame determination step includes: On the plane of the coordinate system where the sensor is located, the initial boundary model of multiple boundaries of the region to be detected is obtained based on the original target data of the current frame; For each initial boundary model, multiple boundary searches are performed within a preset angle range and a preset distance range to determine the boundary parameters of each boundary in the region to be detected. These boundary parameters include confidence levels; and The boundary with the highest confidence level is selected as the feature boundary, and the installation method of the sensor is confirmed based on the correspondence between the angle of the feature boundary and the preset angle range.
4. The method according to claim 2, characterized in that, The multi-frame determination step includes: In response to the single-frame judgment step reaching a preset number of frames, the number of frames within the preset number of frames is counted to determine the sensor's installation method as either front-mounted, side-mounted, or top-mounted; and The installation method of the sensor is determined by comparing the number of frames installed on the front, side, or top with a preset range of frames.
5. The method according to any one of claims 1-4, characterized in that, The sensor has a boundary detection module adapted to various installation methods. The boundary detection of the area to be detected based on the sensor's installation method includes: The boundary detection module adapted to the installation method performs boundary detection on the area to be detected.
6. The method according to any one of claims 1-4, characterized in that, The detection of the area to be detected includes: Obtain the at least two frames of raw target data; Boundary detection processing is performed on each frame of the original target data to obtain the boundary detection results for each frame of the region to be detected; and The boundary detection results of the preset number of frames are subjected to multi-frame comprehensive quantization detection to determine the target boundary detection result of the region to be detected.
7. A boundary detection device, characterized in that, For boundary detection by sensors positioned within the detection area, including: The acquisition module is used to acquire at least two frames of raw target data; An installation method determination module is used to analyze the at least two frames of raw target data to determine the installation method of the sensor, wherein the sensor is installed as a top mount, side mount, or front mount; and The boundary detection module is capable of performing boundary detection on the area to be detected based on the installation method of the sensor.
8. The boundary detection device according to claim 7, characterized in that, The sensor is a radar.
9. A target detection method, applied to detect targets within a specific region having physical boundaries, characterized in that, include: The boundary detection result of the specific region is obtained by applying the boundary detection method of any one of claims 1-6 or the boundary detection device of any one of claims 7-8; and The target is detected based on the boundary detection results.
10. A boundary detection device, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-6 and 9.
11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-6 and 9.
12. A device, characterized in that, include: The device body; and the detection device as described in claim 7 or 8 disposed on the device body; The sensor is used for: boundary / target detection, and / or for communication.
Citation Information
Patent Citations
Road boundary detection method based on vehicle-mounted millimeter wave radar
CN112597839A
Determination of a mounting configuration of a sensor for motor vehicles
WO2021185637A1
