Target detection method, terminal and storage medium

Abstract

The application provides a target detection method, a terminal and a storage medium, and belongs to the field of vehicle-mounted millimeter wave radars.The target detection method comprises the following steps: acquiring distance-Doppler data corresponding to a current frame of echo signals, acquiring target data satisfying preset conditions in terms of distance and Doppler velocity from the distance-Doppler data, and calculating a global noise floor of the current frame of echo signals according to the target data; traversing the distance-Doppler data according to the global noise floor and a segmented detection threshold, selecting point clouds with amplitudes greater than corresponding basic signal-to-noise ratios in each distance interval in the distance dimension to form an initial point cloud set; screening point clouds satisfying target characteristics from the initial point cloud set according to preset target point cloud conditions to form a final point cloud set, and performing target detection according to the final point cloud set.The technical scheme of the application can adapt to different environments, improves the target detection accuracy, avoids missing detection of small targets, and makes the outline of large targets clearer.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of vehicle-mounted millimeter wave radar, and in particular to a target detection method, a terminal and a storage medium. BACKGROUND

[0002] The basic working principle of a millimeter wave radar is to transmit an electromagnetic wave signal and receive a return signal. By processing the return signal, target information of interest in the background is extracted, so as to realize detection of the target by the radar and acquisition of effective information, such as basic information of the target, such as distance, speed, angle, etc. With the development of processing algorithms, more complex target information can also be obtained, such as target acceleration, heading angle, target contour shape, target type judgment, etc. Rich target information relies on complex signal processing and data processing algorithms. Target detection is an important link in radar signal processing, and the quality of the detected target affects the implementation of subsequent complex functions.

[0003] Conventional target detection generally adopts a constant false alarm rate (CFAR) method, such as CA-CFAR, GO-CFAR, OS-CFAR, etc. These methods basically all adopt a neighborhood comparison method, that is, when detecting a target, the signal-to-noise ratio of the current point is compared with the signal-to-noise ratio of the statistical noise in the neighborhood. If the threshold decision condition is met, the point is detected. If the threshold decision condition is not met, the point is excluded. These CFAR methods are widely used. When applied to a vehicle-mounted millimeter wave radar, these methods have certain limitations in some special scenarios, and the detection effect is not ideal, which limits the performance of the radar. For example, when a strong target is next to a small target, the neighborhood detection method can easily cause missed detection, resulting in loss of weak targets, such as bicycle targets next to iron walls, motorcycle targets next to large vehicles, etc. In addition, for large targets, such as large transport trucks, large buses, long trailer trailers, etc., due to their large size, only a part of the target can be detected by the neighborhood method, and the target body cannot be completely detected. Similar effects can also occur for large continuous clutter backgrounds, and the point cloud of the clutter background is not rich and complete. In addition, through the neighborhood detection method, the noise floor formed according to the neighborhood relationship changes with the position of the target in the detection data, that is, the noise floor corresponding to the target in different environments is not fixed. This causes the signal-to-noise ratio of the target to not truly reflect the strength of the target. When the target moves from far to near, the signal-to-noise ratio of the target cannot well present a relationship of inverse fourth power with the distance. In some environments, the signal-to-noise ratio of a target with a small radar cross section (RCS) is greater than that of a target with a large RCS, which is not convenient for direct use of target recognition algorithms, etc.

[0004] The prior art provides a target detection method aiming at the defects in the above conventional target detection method, which can effectively improve the shielding effect of strong targets on weak targets in a multi-target scene. In the implementation of the embodiments of the present application, it is found that the prior art cannot solve the problems of easy missed detection of small targets and incomplete detection of large targets. SUMMARY

[0005] Therefore, the embodiments of the present application provide a target detection method, a terminal and a storage medium to solve the problems of easy missed detection of small targets and incomplete detection of large targets in the prior art target detection technology.

[0006] In a first aspect, the embodiments of the present application provide a target detection method, comprising:

[0007] obtaining distance-Doppler data corresponding to a current frame echo signal, obtaining target data satisfying a preset condition in terms of distance and Doppler velocity from the distance-Doppler data, and calculating a global noise floor of the current frame echo signal according to the target data;

[0008] traversing the distance-Doppler data according to the global noise floor and a segmented detection threshold, selecting point clouds with amplitudes greater than corresponding basic signal-to-noise ratios in each distance interval on the distance dimension to form an initial point cloud set; wherein the basic signal-to-noise ratio corresponding to each distance interval is the sum of the global noise floor and the segmented detection threshold corresponding to the distance interval;

[0009] screening point clouds satisfying target characteristics from the initial point cloud set according to a preset target point cloud condition to form a final point cloud set, and performing target detection according to the final point cloud set.

[0010] In a possible implementation manner, the obtaining of the target data satisfying the preset condition in terms of distance and Doppler velocity from the distance-Doppler data comprises:

[0011] obtaining a target position from the distance-Doppler data, which satisfies a first preset distance condition in terms of distance on the distance dimension and a second preset velocity condition in terms of relative velocity on the Doppler dimension;

[0012] determining a target region according to the target position and preset distance dimension length and Doppler dimension length, and the point cloud data contained in the target region is the target data.

[0013] In a possible implementation manner, the target region comprises a plurality of sub-regions with the same size, and the calculating of the global noise floor of the current frame echo signal according to the target data comprises:

[0014] sequentially calculating the amplitude mean of the point cloud data in each sub-region;

[0015] The amplitude mean value corresponding to the sub-region with the smallest amplitude mean value is taken as the global noise floor of the current frame echo signal.

[0016] In a possible implementation, the target detection method further includes:

[0017] The global noise floor of the N frame historical echo signals including the current frame echo signal is acquired;

[0018] The mean value of the global noise floors of the N frame historical echo signals is calculated, and the calculation result is taken as the final global noise floor of the current frame echo signal.

[0019] In a possible implementation, the target detection method further includes:

[0020] A reference target is determined according to the type of the target to be detected by the radar;

[0021] The segmented detection threshold is generated according to the signal-to-noise ratio characteristic curve of the reference target.

[0022] In a possible implementation, the generating of the segmented detection threshold according to the signal-to-noise ratio characteristic curve of the reference target includes:

[0023] The maximum representable distance range of the radar is divided into a plurality of distance intervals;

[0024] For each distance interval, the signal-to-noise ratio at the farthest distance in the distance interval is acquired according to the signal-to-noise ratio characteristic curve, and the acquired signal-to-noise ratio at the farthest distance is taken as the segmented detection threshold of the distance interval.

[0025] In a possible implementation, the filtering of the point clouds satisfying the target characteristics from the initial point cloud set to form the final point cloud set according to the preset target point cloud condition includes:

[0026] When the number of point clouds in the initial point cloud set is less than or equal to the point cloud number threshold of the final point cloud set, all the point clouds in the initial point cloud set are filtered to form the final point cloud set;

[0027] When the number of point clouds in the initial point cloud set is greater than the point cloud number threshold of the final point cloud set, the number of point clouds whose signal-to-noise ratio and neighborhood point number are high and large is filtered according to the signal-to-noise ratio and neighborhood point number of each point cloud in the initial point cloud set to form the final point cloud set.

[0028] In a possible implementation, the selecting of the point clouds with amplitudes greater than the corresponding basic signal-to-noise ratio in each distance interval in the distance dimension to form the initial point cloud set further includes:

[0029] For the point cloud with an amplitude greater than the basic signal-to-noise ratio, the neighborhood point number of the point cloud is counted and recorded.

[0030] The number of neighborhood points is the number of adjacent points of the point cloud whose amplitudes are greater than the amplitudes of adjacent points in all adjacent points of the point cloud.

[0031] In a second aspect, an embodiment of the present application provides a target detection device, comprising:

[0032] The noise floor obtaining module is configured to obtain distance-Doppler data corresponding to a current frame of echo signals, obtain target data satisfying a preset condition in terms of distance and Doppler velocity from the distance-Doppler data, and calculate a global noise floor of the current frame of echo signals according to the target data.

[0033] The data traversal module is configured to traverse the distance-Doppler data according to the global noise floor and a segment detection threshold, and select point clouds with amplitudes greater than corresponding basic signal-to-noise ratios in each distance interval in the distance dimension to form an initial point cloud set; wherein the basic signal-to-noise ratio corresponding to each distance interval is the sum of the global noise floor and a segment detection threshold corresponding to the distance interval.

[0034] The point cloud screening module is configured to screen point clouds satisfying a target feature from the initial point cloud set according to a preset target point cloud condition to form a final point cloud set, and perform target detection according to the final point cloud set.

[0035] In a third aspect, an embodiment of the present application provides a terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the target detection method as described above when executing the computer program.

[0036] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program, wherein the computer program is executable on a processor to implement the steps of the target detection method as described above.

[0037] The target detection method provided by the embodiment of the present application has the following advantages:

[0038] The application can reduce the influence of targets or clutter on noise floor statistics in global range Doppler data by obtaining distance Doppler data corresponding to the current frame echo signal, obtaining target data satisfying preset conditions from the distance Doppler data, and calculating the global noise floor of the current frame echo signal according to the target data. The global noise floor and the segmented detection threshold are traversed in the distance Doppler data, and the point cloud with an amplitude greater than the corresponding basic signal-to-noise ratio in each distance interval in the distance dimension is selected to form an initial point cloud set, and the basic signal-to-noise ratio corresponding to each distance interval is the sum of the global noise floor and the segmented detection threshold corresponding to the distance interval, so that the point cloud with a low signal-to-noise ratio can be screened out, thereby obtaining high-quality and high-precision point cloud representing target characteristics. The point cloud satisfying the target characteristics is selected from the initial point cloud set according to the preset target point cloud condition to form a final point cloud set, and target detection is performed according to the final point cloud set, thereby reducing the probability of missing small targets, and the point cloud of large targets is also more abundant, which can better reflect the outline of the target. The technical scheme of the application can adapt to different environments and improve the target detection accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0040] Figure 1 is an application scenario diagram of a target detection method provided by an embodiment of the present application;

[0041] Figure 2 is a flowchart of the implementation of a target detection method provided by an embodiment of the present application;

[0042] Figure 3 is a global range Doppler data and noise floor statistical sub-region position schematic diagram of a target detection method provided by an embodiment of the present application;

[0043] Figure 4 is a detection threshold setting schematic diagram according to distance segmentation of a target detection method provided by an embodiment of the present application;

[0044] Figure 5 is a neighborhood point number statistical schematic diagram of a target detection method provided by an embodiment of the present application;

[0045] Figure 6 is a schematic diagram of a target detection device provided by an embodiment of the present application;

[0046] Figure 7 is a schematic diagram of a terminal provided by an embodiment of the present application. Detailed Implementation

[0047] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0048] To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiments will be described below in conjunction with the accompanying drawings.

[0049] See Figure 1 The illustration shows an application scenario of a target detection method provided by an embodiment of the present invention. This embodiment of the present invention pertains to the field of vehicle-mounted radar. The radar detection range covers the road and surrounding environment where the vehicle is located. Large and small targets exist within the radar detection range. In real-world applications, large targets include large objects such as trucks, buses, and trees, while small targets include smaller objects such as bicycles, trams, motorcycles, and pedestrians.

[0050] In existing technologies, the presence of smaller targets near large targets can easily lead to a shielding effect. For example, if there is a bicycle next to a truck, the truck may block the bicycle, making it difficult for the radar to detect the bicycle and causing it to be missed. At the same time, large targets such as trucks and buses are large in size, which can easily lead to incomplete target detection.

[0051] The target detection method proposed in this invention can effectively reveal the outline of large targets while ensuring that small targets are not missed, thus improving detection performance and increasing target detection accuracy. This invention is particularly suitable for targets with a large RCS or for scenarios where the target is in a large RCS, such as when there are metal fences or railings along the roadside, as these fences or railings have a strong ability to reflect radar waves, resulting in good radar target detection performance.

[0052] In addition Figure 1 In addition to the scenario of vehicles driving on the road, the target detection method provided by this invention is also applicable to scenarios such as vehicles parked on the side of the road or in parking lots.

[0053] See Figure 2 The diagram illustrates a flowchart of a target detection method provided by an embodiment of the present invention, which includes the following steps:

[0054] S201. Obtain the range Doppler data corresponding to the echo signal of the current frame, obtain the target data from the range Doppler data that meets the preset conditions for range and Doppler velocity, and calculate the global noise floor of the echo signal of the current frame based on the target data.

[0055] In the embodiment of the present application, the radar transmits electromagnetic wave signals and receives echo signals, the range-doppler data includes global data that can be detected by the radar, and the most representative region needs to be selected from the global range-doppler data as target data to calculate the global noise floor.

[0056] In a possible implementation, the target data satisfying a preset condition in terms of range and doppler velocity is obtained from the range-doppler data, including:

[0057] a target position satisfying a first preset distance condition in terms of distance from the radar in the range dimension and a second preset velocity condition in terms of relative velocity from the radar in the doppler dimension is obtained from the range-doppler data.

[0058] a target region is determined according to the target position and preset range dimension length and doppler dimension length, and point cloud data contained in the target region is the target data.

[0059] In the embodiment of the present application, the position farthest from the radar in the range dimension and having a relatively large relative velocity from the radar in the doppler dimension is selected as the target position from the range-doppler data, and the statistics and calculation of the global noise floor are based on the global range-doppler data. In the range-doppler data, the energy of the target and clutter is relatively strong at a near distance. If this part of data segment with relatively strong energy is used to statistically calculate the noise floor, the noise floor obtained by the statistics will not be the real noise floor, but will be higher than the real noise floor. The signal-to-noise ratio of the target detected by using such a noise floor will also not be the real signal-to-noise ratio of the target. This will lead to the fact that the signal-to-noise ratio characteristics of the target at different distance segments do not conform to the radar equation, which is not convenient for target identification or other algorithms to use and refer to the signal-to-noise ratio information of the target. The noise floor obtained by the statistics of the target data farthest in the range dimension and having a relatively large relative velocity is close to the real noise floor, and the influence of the relatively strong target or clutter is avoided. The range dimension size and the doppler dimension size of the target region can be adjusted according to the size of the range-doppler data.

[0060] In a possible implementation, the target data satisfying a preset condition in terms of range and doppler velocity is obtained from the range-doppler data, including:

[0061] a target position satisfying a first preset distance condition in terms of distance from the radar in the range dimension is obtained from the range-doppler data.

[0062] a target region is determined according to the target position and preset range dimension length and doppler dimension length, and point cloud data contained in the target region is the target data.

[0063] In the embodiment, it is suitable for the case that the targets in the detection range of the radar are in a stationary state or the relative velocities of the targets from the radar are relatively small.

[0064] In a possible implementation, the target data satisfying the preset condition in distance and Doppler velocity is obtained from the range-Doppler data further includes:

[0065] The target position satisfying a second preset velocity condition in Doppler velocity relative to the radar is obtained from the range-Doppler data;

[0066] The target region is determined according to the target position and the preset distance dimension length and Doppler velocity length, and the point cloud data contained in the target region is the target data.

[0067] In the embodiment, it is suitable for the case that the radar detection range is small and the relative velocities of the targets relative to the radar are greatly different.

[0068] In a possible implementation, the target region includes a plurality of sub-regions with the same size; and the global noise floor of the current frame echo signal is calculated according to the target data, including:

[0069] The amplitude mean values of the point cloud data in each sub-region are sequentially calculated;

[0070] The amplitude mean value corresponding to the sub-region with the minimum amplitude mean value is taken as the global noise floor of the current frame echo signal.

[0071] Generally, the distance and Doppler of a target or clutter do not simultaneously span a large range, even if there is a target or clutter in a sub-region, the average value of the sub-region may be raised, but there is no target or clutter in the remaining sub-regions, and the amplitude mean value corresponding to the sub-region with the minimum amplitude mean value is taken as the global noise floor of the current frame echo signal, which can reduce the influence of the target or clutter on the absolute noise floor and truly reflect the noise characteristics of the range-Doppler data.

[0072] Referring to Figure 3 , a schematic diagram of global range-Doppler data and noise floor statistical sub-region positions of a target detection method provided by an embodiment of the present application is shown, assuming that the size of the global range-Doppler data is MxN, M is the total number of Doppler units, N is the total number of distance units, and the region satisfying the preset condition in distance and relative velocity in the global range-Doppler data is selected as the target region, for example, the distance is near 15 / 16 of the total distance unit number of the range-Doppler, and the Doppler is near 1 / 2 of the total Doppler unit number of the range-Doppler, and 8 sub-regions shown in Figure 3 are selected as the target region for statistical noise floor, the size of each sub-region in the distance direction can be selected as 1 / 32 of the total distance unit number, and the size of each sub-region in the Doppler direction can be selected as 1 / 32 of the total distance unit number, the amplitude mean values of each piece of data in the sub-regions ①-⑧ are respectively calculated, and the minimum value of the 8 amplitude mean values is selected as the global noise floor of the current frame echo signal.

[0073] In a possible implementation, the target detection method further comprises:

[0074] obtaining a global noise floor of N frames of historical echo signals including the current frame echo signal;

[0075] calculating a mean value of the global noise floors of the N frames of historical echo signals, and taking the calculation result as the final global noise floor of the current frame echo signal.

[0076] In the embodiment of the application, the statistical mean value of the global noise floors of the N frames of historical echo signals can further reduce accidental fluctuations of a certain frame noise floor, so that the noise floor is more stable and close to the true level.

[0077] In a possible implementation, the target detection method further comprises:

[0078] determining a reference target according to a type of target to be detected by the radar;

[0079] generating a segmented detection threshold according to a signal-to-noise ratio characteristic curve of the reference target.

[0080] In the embodiment of the application, a reference target is selected according to the type of target to be detected by the vehicle-mounted millimeter wave radar, and a small target is generally selected as the reference target. The RCS of the reference target is smaller than that of the target to be detected, for example, a two-wheeled vehicle with an RCS of 5 dBsm is selected as the reference target. If the RCS of the target to be detected is very small, a target with a smaller RCS can be selected as the reference target, or the detection distance can be reduced, so that the small target is not missed. Selecting the reference target according to the type of target to be detected by the vehicle-mounted millimeter wave radar can detect large targets and small targets to be detected.

[0081] In the maximum distance range of the radar, the signal-to-noise ratio characteristic curve of the target at different distances can be obtained by combining actual measurement and simulation. Figure 4 FIG. 1 shows a schematic diagram of setting a detection threshold according to a distance in a target detection method according to an embodiment of the application, and a two-wheeled vehicle with an RCS of 5 dBsm is taken as a reference target to obtain a signal-to-noise ratio characteristic curve of the reference target at different distances.

[0082] In a possible implementation, generating a segmented detection threshold according to the signal-to-noise ratio characteristic curve of the reference target comprises:

[0083] dividing the maximum representable distance range of the radar into a plurality of distance intervals;

[0084] for each distance interval, obtaining a signal-to-noise ratio at the farthest distance in the distance interval according to the signal-to-noise ratio characteristic curve, and taking the obtained signal-to-noise ratio at the farthest distance as the segmented detection threshold of the distance interval.

[0085] In the embodiment of the present application, the adoption of the segmented detection threshold is another feature that distinguishes the present application from the conventional CFAR detection method, as shown in Figure 4 Taking the target SNR threshold at the farthest detectable distance of 12dB as an example, the SNR intensity curve of a 5dBsm target at different distances can be obtained. Considering the possible fluctuation, the SNR characteristic curve is lowered by 4dB, and the maximum representable distance range of the radar is segmented at equal intervals or unequal intervals, for example, every 8m distance is taken as a segment, and the SNR threshold at the farthest distance in each distance segment is taken as the target detection threshold of the distance segment in the manner shown in Figure 4

[0086] In the near distance segment, there is a high-pass filter to filter out the direct current component generated by the coupling between the radar transceiver channels, so that the amplitude of the target or clutter is attenuated by 20-30dB. Therefore, in actual application, the influence of the high-pass filter needs to be considered when setting the detection threshold in the near distance segment, and the SNR threshold in these distance segments is adjusted according to the attenuation characteristics of the filter, for example, the detection threshold at a distance within 20m needs to be lowered by 20-30dB based on the SNR threshold.

[0087] Figure 4 It can be seen that the SNR intensity of the target in different distance segments is different, and if the same detection threshold is used throughout the distance, many high-quality points cannot be selected, and the noise influence caused by some clutters cannot be filtered out. In the embodiment of the present application, the segmented detection threshold is set according to the SNR characteristic curve of the reference target, and the points in each distance segment that can better represent the SNR characteristics of the target are selected, so that more abundant points can be obtained.

[0088] S202, traverse the range Doppler data according to the global noise floor and the segmented detection threshold, and select point clouds with amplitudes greater than corresponding basic SNRs in each distance interval in the range dimension to form an initial point cloud set; wherein the basic SNR corresponding to each distance interval is the sum of the global noise floor and the segmented detection threshold corresponding to the distance interval.

[0089] In the embodiment of the present application, the lower the SNR of a point or the smaller the number of neighborhood points, the farther the point is from the peak vertex, and the closer it is to the envelope sidelobe or the valley bottom, and the smaller the contribution to target detection. According to the global noise floor and the segmented detection threshold, the global range Doppler data is traversed to obtain an initial point cloud set, so that points with low SNR and small number of neighborhood points are filtered out, and point clouds meeting the requirements are obtained, which are high in quality and good in accuracy, and can better represent the characteristics of the target.

[0090] In one possible implementation, selecting point clouds with amplitudes greater than corresponding basic SNRs in each distance interval in the range dimension to form an initial point cloud set further includes: ​

[0091] For the point cloud with the amplitude greater than the base signal-to-noise ratio, the number of neighborhood points of the point cloud is counted and recorded, wherein the number of neighborhood points is the number of neighboring points of the point cloud in all neighboring points of the point cloud, and the amplitude of the neighboring point of the point cloud is greater than the amplitude of the neighboring point.

[0092] Referring to Figure 5 which shows a neighborhood point number counting schematic diagram of a target detection method provided by an embodiment of the present application, the so-called counting of the number of neighborhood points is to compare the signal-to-noise ratio size relationship of the current point and the surrounding neighboring points, assuming that the to-be-detected point is located at a non-edge position of the global range-doppler data, then there are 8 neighboring points around each point, for example Figure 5 the position of the 5th point in the middle, if the signal-to-noise ratio of the 5th point is greater than that of all the surrounding points, then the number of neighborhood points of the point is 8. Similarly, if Figure 5 the signal-to-noise ratio of the 5th point in the middle is only greater than that of 7 points among the surrounding neighboring points, then the number of neighborhood points of the point is 7, if Figure 5 the signal-to-noise ratio of the 5th point in the middle is only greater than that of 6 points among the surrounding neighboring points, then the number of neighborhood points of the point is 6, and so on. The points with the number of neighborhood points being 8 are all peak vertices of the target or clutter envelope, the smaller the number of neighborhood points is, the farther the point is away from the peak vertex, and the closer the point is to the envelope sidelobe or valley bottom. For the points located at the edge of the global range-doppler data, the neighboring points of each point are at most 5 or 3.

[0093] S203, filtering the point clouds satisfying the target characteristics from the initial point cloud set according to a preset target point cloud condition to form a final point cloud set, and performing target detection according to the final point cloud set.

[0094] In the embodiment of the present application, since the RCS of the selected reference target type is generally small, the segmented detection threshold is low, and therefore the total number of points in the initial point cloud set is large. In order to prevent too many points from occupying too many system processing and storage resources, the initial point cloud set needs to be filtered.

[0095] In a possible implementation manner, filtering the point clouds satisfying the target characteristics from the initial point cloud set according to a preset target point cloud condition to form a final point cloud set comprises:

[0096] When the number of point clouds in the initial point cloud set is less than or equal to the point cloud number threshold of the final point cloud set, filtering all the point clouds in the initial point cloud set to form the final point cloud set;

[0097] When the number of point clouds in the initial point cloud set is greater than the point cloud number threshold of the final point cloud set, filtering the point clouds with high signal-to-noise ratio and large number of neighborhood points to form the final point cloud set, wherein the number of the point clouds is the point cloud number threshold.

[0098] In the embodiment of the present application, when the number of point clouds in the initial point cloud set is greater than the point cloud number threshold of the final point cloud set, the initial point cloud set is sorted in a joint descending order of the number of neighborhood points and the signal-to-noise ratio, and the point with a larger number of neighborhood points is arranged in front. These points are all peak points of the target or points on the main lobe, and are the part of the target point cloud with the highest precision and the largest amount of information. When the number of neighborhood points is the same, the point is arranged in descending order according to the signal-to-noise ratio. The result is that all the peak points of the target in the initial point cloud set are selected. Among these peak points, some are targets at a distance, and the signal-to-noise ratio is small. If only the conventional signal-to-noise ratio sorting method is used, the target with a small signal-to-noise ratio at a distance is discarded, which easily causes the target to be missed and reduces the effective detection range of the target, which brings greater risk to some advanced auxiliary driving or automatic driving systems.

[0099] The initial point cloud set is sorted in a joint descending order of the number of neighborhood points and the signal-to-noise ratio. After sorting, the point with a large number of neighborhood points and a high signal-to-noise ratio is arranged in front, and the point with a small number of neighborhood points and a low signal-to-noise ratio is arranged in back. The main envelope of the target or clutter is selected, the probability of missing small targets is reduced, and the point cloud of large targets is more abundant, which better reflects the outline of the target.

[0100] The vehicle-mounted millimeter wave radar further processes the final point cloud set, such as AoA processing to obtain the azimuth, pitch angle and static clutter of the target, separation of moving targets, target clustering, track tracking, target recognition, drivable area judgment and emergency braking and other complex processing, and can also cooperate with the vehicle domain controller to realize the functions of powerful advanced driving assistance or automatic driving.

[0101] In the embodiment, the distance-Doppler data corresponding to the current frame echo signal is obtained, target data satisfying a preset condition in distance and Doppler velocity is obtained from the distance-Doppler data, the global noise floor of the current frame echo signal is calculated according to the target data, the influence of the target or clutter on the noise floor statistics in the global distance-Doppler data is reduced; the distance-Doppler data is traversed according to the global noise floor and the segmented detection threshold, and the point cloud with an amplitude greater than the corresponding basic signal-to-noise ratio in each distance interval in the distance dimension is selected to form an initial point cloud set; wherein the basic signal-to-noise ratio corresponding to each distance interval is the sum of the global noise floor and the segmented detection threshold corresponding to the distance interval, and the point cloud with a low signal-to-noise ratio is screened out, thereby obtaining high-quality and high-precision point clouds representing target characteristics; the point cloud satisfying the target characteristics is screened from the initial point cloud set according to a preset target point cloud condition to form a final point cloud set, and target detection is performed according to the final point cloud set, thereby reducing the probability of missing small targets, and the point cloud of large targets is more abundant, which better reflects the outline of the target. The technical scheme of the present application can adapt to different environments and improve the target detection accuracy.

[0102] It should be understood that the size of the serial number of each step in the above embodiment does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute a limitation on the implementation process of the embodiment of the application.

[0103] The following is an apparatus embodiment of the application. For details not described herein, reference can be made to the above-mentioned method embodiments.

[0104] Figure 6 A structural schematic diagram of a target detection apparatus provided by an embodiment of the application is shown. For ease of illustration, only parts related to the embodiment of the application are shown, such as Figure 6 The target detection apparatus includes a noise floor acquisition module 601, a data traversal module 602, and a point cloud screening module 603.

[0105] The noise floor acquisition module 601 is configured to acquire distance-Doppler data corresponding to a current frame of echo signals, acquire target data satisfying a preset condition in terms of distance and Doppler velocity from the distance-Doppler data, and calculate a global noise floor of the current frame of echo signals according to the target data.

[0106] The data traversal module 602 is configured to traverse the distance-Doppler data according to the global noise floor and a segmented detection threshold, and select point clouds with amplitudes greater than a corresponding basic signal-to-noise ratio in each distance interval in the distance dimension to form an initial point cloud set; wherein the basic signal-to-noise ratio corresponding to each distance interval is the sum of the global noise floor and a segmented detection threshold corresponding to the distance interval.

[0107] The point cloud screening module 603 is configured to screen point clouds satisfying a target characteristic from the initial point cloud set according to a preset target point cloud condition to form a final point cloud set, and perform target detection according to the final point cloud set.

[0108] In a possible implementation manner, the noise floor acquisition module 601 is specifically configured to acquire, from the distance-Doppler data, a target position satisfying a first preset distance condition in terms of distance from the radar and a second preset velocity condition in terms of relative velocity of the radar;

[0109] determine a target region according to the target position and a preset distance dimension length and Doppler dimension length, and the point cloud data contained in the target region is the target data.

[0110] In a possible implementation manner, the noise floor acquisition module 601 is specifically configured to sequentially calculate the amplitude mean of the point cloud data in each sub-region.

[0111] The amplitude mean of the sub-region corresponding to the smallest amplitude mean is taken as the global noise floor of the current frame of echo signals.

[0112] In a possible implementation, the noise floor obtaining module 601 is specifically configured to: obtain a global noise floor of N frames of historical echo signals including the current frame echo signal;

[0113] calculate a mean value of the global noise floor of the N frames of historical echo signals, and take the calculation result as the final global noise floor of the current frame echo signal.

[0114] In a possible implementation, the noise floor obtaining module 601 is specifically configured to: determine a reference target according to a target type to be detected by the radar;

[0115] generate a segmented detection threshold according to a signal-to-noise ratio characteristic curve of the reference target.

[0116] In a possible implementation, the noise floor obtaining module 601 is specifically further configured to: divide a maximum representable distance range of the radar into a plurality of distance intervals;

[0117] For each distance interval, obtain a signal-to-noise ratio at a farthest distance in the distance interval according to the signal-to-noise ratio characteristic curve, and take the obtained signal-to-noise ratio at the farthest distance as a segmented detection threshold of the distance interval.

[0118] In a possible implementation, the data traversal module 602 is specifically configured to: traverse the range-Doppler data according to the global noise floor and the segmented detection threshold, and select point clouds with amplitudes greater than corresponding basic signal-to-noise ratios in each distance interval in the range dimension to form an initial point cloud set;

[0119] wherein the basic signal-to-noise ratio corresponding to each distance interval is a sum of the global noise floor and the segmented detection threshold corresponding to the distance interval.

[0120] In a possible implementation, the data traversal module 602 is specifically configured to: when the number of point clouds in the initial point cloud set is less than or equal to a point cloud number threshold of the final point cloud set, filter all point clouds in the initial point cloud set to form the final point cloud set;

[0121] When the number of point clouds in the initial point cloud set is greater than the point cloud number threshold of the final point cloud set, filter point clouds with high signal-to-noise ratios and large numbers of neighbor points to form the final point cloud set.

[0122] In a possible implementation, the data traversal module 602 is specifically further configured to: for the point clouds with amplitudes greater than the basic signal-to-noise ratio, count and record the number of neighbor points of the point clouds;

[0123] wherein the number of neighbor points is the number of neighbor points with amplitudes greater than the amplitude of the point cloud among all neighbor points of the point cloud.

[0124] In a possible implementation, the point cloud screening module 603 is specifically configured to: screen point clouds satisfying a target characteristic from the initial point cloud set according to a preset target point cloud condition to form a final point cloud set, and perform target detection according to the final point cloud set.

[0125] In a possible implementation, the point cloud screening module 603 is specifically further configured to: perform further subsequent processing on the final point cloud set by the vehicle-mounted millimeter wave radar, such as AoA processing to obtain the azimuth, pitch angle and static clutter of the target, separation of the moving target, target clustering, track tracking, target identification, drivable area judgment and emergency braking and other complex processing, and the vehicle-mounted millimeter wave radar can also cooperate with a vehicle domain controller to implement a powerful advanced driving assistance or automatic driving function.

[0126] In this embodiment, the distance-Doppler data corresponding to the current frame echo signal is obtained, target data satisfying a preset condition in terms of distance and Doppler velocity is obtained from the distance-Doppler data, the global noise floor of the current frame echo signal is calculated according to the target data, the influence of the target or clutter on the noise floor statistics in the global distance-Doppler data can be reduced, the point clouds with amplitudes greater than the corresponding basic signal-to-noise ratio in each distance interval in the distance dimension are selected to form an initial point cloud set by traversing the distance-Doppler data according to the global noise floor and the segmented detection threshold, the basic signal-to-noise ratio corresponding to each distance interval is the sum of the global noise floor and the segmented detection threshold corresponding to the distance interval, the point clouds with low signal-to-noise ratio can be screened out, so that high-quality and high-precision point clouds representing target characteristics are obtained, the point clouds satisfying the target characteristics are screened from the initial point cloud set according to a preset target point cloud condition to form a final point cloud set, and target detection is performed according to the final point cloud set, so that the probability of missing small targets is reduced, the point clouds of large targets are also more abundant, and the outline of the target can be better represented. The technical scheme of the present application can adapt to different environments and improve the target detection accuracy.

[0127] Referring to Figure 7 , a schematic diagram of a terminal is shown. As Figure 7 shown, the terminal 7 of this embodiment includes a processor 70, a memory 71, and a computer program 72 stored in the memory 71 and executable on the processor 70. The processor 70 implements the steps in each of the above target detection method embodiments when executing the computer program 72, for example Figure 2 steps S201 to S204. Alternatively, the processor 70 implements the functions of each module / unit in each of the above device embodiments when executing the computer program 72, for example Figure 6 the functions of the modules 601 to 603 shown.

[0128] For example, the computer program 72 can be divided into one or more modules / units, which are stored in the memory 71 and executed by the processor 70 to complete the present application. The one or more modules / units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program 72 in the terminal 7. For example, the computer program 72 can be divided into the following modules: Figure 6 The modules 601 to 603 shown.

[0129] The terminal 7 can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing devices. The terminal 7 can include, but is not limited to, the processor 70, the memory 71. Those skilled in the art can understand that the terminal 7 can include more or fewer components than those shown, or combine some components, or different components, for example, the terminal can also include an input / output device, a network access device, a bus, etc. Figure 7 The terminal 7 shown is only an example and does not constitute a limitation on the terminal 7, and can include more or fewer components than those shown, or combine some components, or different components, for example, the terminal can also include an input / output device, a network access device, a bus, etc.

[0130] The processor 70 can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0131] The memory 71 can be an internal storage unit of the terminal 7, such as a hard disk or a memory of the terminal 7. The memory 71 can also be an external storage device of the terminal 7, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory 71 can include both the internal storage unit and the external storage device of the terminal 7. The memory 71 is used to store the computer program and other programs and data required by the terminal. The memory 71 can also be used to temporarily store data that has been output or will be output.

[0132] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is exemplified, and in actual application, the above-mentioned functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be realized in the form of hardware or software. In addition, the specific names of each functional unit and module are only for easy distinction, and do not limit the protection scope of the present application. The specific working process of the units and modules in the above system can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0133] In the above embodiments, the description of each embodiment has its own emphasis, and the parts not described or recorded in detail in a certain embodiment can be referred to the related description of other embodiments.

[0134] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0135] In the embodiments provided by the present application, it should be understood that the disclosed apparatus / terminal and method can be implemented by other ways. For example, the above-mentioned apparatus / terminal embodiments are only schematic, and the division of the modules or units is only a logical function division, and there can be another division way in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0136] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0137] In addition, each function unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software function unit.

[0138] The integrated module / unit, if realized in the form of a software function unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer-readable storage medium. When the processor executes the computer program, the steps of each target detection method embodiment can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, an executable file, or some intermediate form, etc. The computer-readable medium can include any entity or device capable of carrying the computer program code, a recording medium, a U disk, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc. It should be noted that the contents included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0139] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present application, and should be included in the protection scope of the present application.

Claims

1. A target detection method, characterized in that, include: Acquire the range Doppler data corresponding to the current frame echo signal, obtain target data from the range Doppler data that satisfies the preset conditions for range and Doppler velocity, and calculate the global noise floor of the current frame echo signal based on the target data; Based on the global noise floor and the segmented detection threshold, the range Doppler data is traversed, and point clouds with amplitudes greater than the corresponding basic signal-to-noise ratios within each distance interval in the distance dimension are selected to form an initial point cloud set; wherein, the basic signal-to-noise ratio corresponding to each distance interval is the sum of the global noise floor and the segmented detection threshold corresponding to that distance interval; Based on preset target point cloud conditions, point clouds that meet the target characteristics are selected from the initial point cloud set to form a final point cloud set, and target detection is performed based on the final point cloud set; The step of selecting point clouds that meet the target characteristics from the initial point cloud set according to the preset target point cloud conditions to form the final point cloud set includes: When the number of point clouds in the initial point cloud set is less than or equal to the number of point clouds in the final point cloud set, all point clouds in the initial point cloud set are selected to form the final point cloud set. When the number of point clouds in the initial point cloud set is greater than the number of point clouds in the final point cloud set, the final point cloud set is formed by selecting point clouds with high signal-to-noise ratio and large number of neighboring points based on the signal-to-noise ratio and the number of neighboring points of each point cloud in the initial point cloud set.

2. The target detection method according to claim 1, characterized in that, The target data obtained from the range-Doppler data that satisfies the preset conditions for distance and Doppler velocity includes: The target position is obtained from the range Doppler data, where the distance between the target and the radar in the range dimension satisfies a first preset distance condition, and the relative velocity between the target and the radar in the Doppler dimension satisfies a second preset velocity condition. The target region is determined based on the target location and the preset distance dimension and Doppler dimension, and the point cloud data contained within the target region is the target data.

3. The target detection method according to claim 2, characterized in that, The target region includes multiple sub-regions of the same size; the calculation of the global noise floor of the current frame echo signal based on the target data includes: Calculate the mean amplitude of the point cloud data in each sub-region sequentially; The amplitude mean of the sub-region with the smallest amplitude mean is taken as the global noise floor of the echo signal of the current frame.

4. The target detection method according to claim 1, characterized in that, The target detection method further includes: Obtain the global noise floor of N consecutive frames of historical echo signals, including the echo signal of the current frame; Calculate the mean global noise floor of the N historical echo signals, and use the calculation result as the final global noise floor of the current frame echo signal.

5. The target detection method according to claim 1, characterized in that, The target detection method further includes: Determine the reference target based on the type of target to be detected by the radar; The segmented detection threshold is generated based on the signal-to-noise ratio characteristic curve of the reference target.

6. The target detection method according to claim 5, characterized in that, The step of generating the segmented detection threshold based on the signal-to-noise ratio characteristic curve of the reference target includes: Divide the maximum representable range of the radar into multiple range intervals; For each distance interval, the signal-to-noise ratio at the farthest distance within that interval is obtained based on the signal-to-noise ratio characteristic curve, and the obtained signal-to-noise ratio at the farthest distance is used as the segmented detection threshold for that distance interval.

7. The target detection method according to claim 1, characterized in that, The step of selecting point clouds whose amplitude is greater than the corresponding basic signal-to-noise ratio within each distance interval of the distance dimension to form the initial point cloud set also includes: For point clouds with amplitudes greater than the base signal-to-noise ratio, count and record the number of neighborhood points of the point cloud; The number of neighboring points is the number of neighboring points in the point cloud whose amplitude is greater than that of the neighboring points.

8. A terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the target detection method as described in any one of claims 1 to 7.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the target detection method as described in any one of claims 1 to 7.

Images