Recognition method and device based on trace point set, storage medium and electronic equipment

Abstract

The invention discloses an identification method and device based on a trace point set, a storage medium and electronic equipment. The method comprises the following steps: acquiring driving state information of a vehicle object and a candidate trace point set corresponding to a current frame; acquiring region boundary description information of a region of interest matched with the current frame under the condition that the driving state information of the self-vehicle object meets a target measuring point condition and the first trace point quantity of the first trace point object is greater than a first quantity threshold value; determining the plurality of candidate trace point objects in the region of interest as a plurality of second trace point objects according to the position information corresponding to the plurality of candidate trace point objects; and executing object recognition operation according to a second trace point set comprising a plurality of second trace point objects. According to the invention, the technical problem of inaccurate identification result based on the trace point set in the prior art is solved.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving, and more specifically, to a point trace set-based recognition method, device, storage medium, and electronic device. Background Art

[0002] In scenes with more complex road environments, there are many real targets and multipath reflection points. When the number of output points is fixed, short-range targets that are not of interest will occupy the space of medium- and long-range targets of interest, resulting in the inability to output medium- and long-range targets normally, ultimately leading to inaccurate object detection based on radar signals.

[0003] At present, the method of directly relaxing the limit on the number of final output points is mostly used to release more mid- and long-range points. However, the method of relaxing the overall output point will increase the time consumption and memory space occupation. For automotive millimeter-wave radars with limited memory and frame length, it is easy to cause the problem of too long frame period, which will also affect the overall performance of the radar. In other words, there is a technical problem in the prior art that the output point set is inaccurate, resulting in inaccurate object recognition results based on point traces. Summary of the invention

[0004] The embodiments of the present application provide a recognition method, device, storage medium and electronic device based on a point trace set, so as to at least solve the technical problem that the recognition method based on a point trace set in the related art is inaccurate.

[0005] According to one aspect of an embodiment of the present application, a recognition method based on a point trace set is provided, comprising: obtaining driving state information of a vehicle object and a candidate point trace set corresponding to a current frame, wherein the candidate point trace set is obtained through a first screening operation based on a first point trace set collected by the vehicle in the current frame, and the first point trace set includes a plurality of first point trace objects detected by constant false alarm; when the driving state information of the vehicle object meets a target measuring point condition and the number of first points of the first point trace object is greater than a first quantity threshold, obtaining area boundary description information of an area of ​​interest matching the current frame; according to the position information corresponding to each of the plurality of candidate point trace objects in the candidate point trace set, determining the plurality of candidate point trace objects located in the area of ​​interest as a plurality of second point trace objects, wherein the number of second points of the plurality of second point trace objects is less than or equal to the first quantity threshold; and performing an object recognition operation based on the second point trace set including the plurality of second point trace objects.

[0006] According to another aspect of an embodiment of the present application, a recognition device based on a point trace set is also provided, including: a first acquisition unit, used to acquire driving state information of a self-vehicle object and a candidate point trace set corresponding to a current frame, wherein the candidate point trace set is obtained through a first screening operation based on the first point trace set collected by the self-vehicle in the current frame, and the first point trace set includes a plurality of first point trace objects detected by constant false alarm; a second acquisition unit, used to acquire area boundary description information of an area of ​​interest matching the current frame when the driving state information of the self-vehicle object meets the target measurement point condition and the number of first points of the first point trace object is greater than a first quantity threshold; a determination unit, used to determine the plurality of candidate point trace objects located in the area of ​​interest as a plurality of second point trace objects according to the position information corresponding to each of the plurality of candidate point trace objects in the candidate point trace set, wherein the number of second points of the plurality of second point trace objects is less than or equal to the first quantity threshold; and a recognition unit, used to perform an object recognition operation based on a second point trace set including the plurality of second point trace objects.

[0007] As an optional scheme, the above-mentioned determination unit includes at least one of the following: a first determination module, which is used to determine a plurality of second point trace objects from a plurality of the above-mentioned candidate point trace objects based on a comparison result between the plane coordinates of the above-mentioned candidate point trace objects and the plane coordinates of at least one of the above-mentioned area boundary points when the above-mentioned area boundary description information includes the plane coordinates of each of the at least one area boundary points; a second determination module, which is used to determine a plurality of second point trace objects from a plurality of the above-mentioned candidate point trace objects based on a comparison result between the azimuth index identifier of the above-mentioned candidate point trace object and the reference distance unit index identifier included in the above-mentioned boundary point description array when the above-mentioned area boundary description information includes a boundary point description array corresponding to at least one area boundary point, wherein the above-mentioned boundary point description array is used to indicate the correspondence between the boundary distance of the boundary point and the azimuth index.

[0008] As an optional scheme, the above-mentioned second acquisition unit includes at least one of the following: a first acquisition module, which is used to determine the above-mentioned area boundary description information according to the position corresponding to the above-mentioned target enclosure object when the driving environment of the above-mentioned self-vehicle object includes the target enclosure object; a second acquisition module, which is used to determine the area boundary description information of the area of ​​interest matching the above-mentioned current frame according to the historical boundary description information corresponding to at least one historical frame when the driving environment of the above-mentioned self-vehicle object does not include the target enclosure object.

[0009] As an optional scheme, the above-mentioned second determination module includes: traversing the above-mentioned multiple candidate point trace objects, and determining the reference distance unit index identifier that matches the above-mentioned azimuth index identifier from the above-mentioned boundary point description array according to the azimuth index identifier of the above-mentioned candidate point trace object currently traversed; determining the reference azimuth index identifier corresponding to the above-mentioned reference distance unit index identifier according to the above-mentioned boundary point description array; when the above-mentioned azimuth index identifier of the above-mentioned candidate point trace object matches the above-mentioned reference azimuth index identifier, determining the above-mentioned candidate point trace object currently traversed as the above-mentioned second point trace object.

[0010] As an optional solution, the above-mentioned first acquisition unit is used for at least one of the following: obtaining the road shape information of the above-mentioned vehicle object; obtaining a point trace processing identifier that matches the historical point trace set collected in the previous frame, wherein the above-mentioned point trace processing identifier is used to indicate the processing method of the above-mentioned first point trace set in the above-mentioned current frame; obtaining the number of historical point traces of the historical point trace object included in the historical point trace set.

[0011] As an optional scheme, the above-mentioned recognition device based on point trace set also includes: a condition detection unit, which is used to determine that the above-mentioned driving state information of the above-mentioned vehicle object does not meet the above-mentioned target measurement point condition when the above-mentioned road shape information indicates that the above-mentioned vehicle is in a non-straight driving state; when the above-mentioned road shape information indicates that the above-mentioned vehicle is in a straight driving state and the above-mentioned point trace processing identifier is a first value, determine that the above-mentioned driving state information of the above-mentioned vehicle object meets the above-mentioned target measurement point condition; when the above-mentioned road shape information indicates that the above-mentioned vehicle is in a straight driving state, and the weighted sum of the above-mentioned first point trace number and the above-mentioned historical point trace number is greater than a reference number threshold, determine that the above-mentioned driving state information of the above-mentioned vehicle object meets the above-mentioned target measurement point condition.

[0012] As an optional scheme, the above-mentioned second determination module also includes at least one of the following: when the ratio between the number of reference points of the reference point trace object and the second number of second points of the above-mentioned second point trace object is greater than a first threshold, the point trace processing identifier of the above-mentioned current frame is determined as the first value, wherein the above-mentioned reference point trace object is a point trace object outside the boundary of the area of ​​interest; when the ratio between the above-mentioned reference point trace number of the above-mentioned reference point trace object and the above-mentioned second number of points of the above-mentioned second point trace object is greater than a second threshold, the above-mentioned point trace processing identifier of the above-mentioned current frame is determined as the above-mentioned first value.

[0013] As an optional scheme, the above-mentioned second determination module is also used to: modify the value of the inter-frame counter matching the current frame to a first count value, wherein the above-mentioned first count value is greater than 0, and the value of the above-mentioned inter-frame counter is used to indicate the number of frames for performing the above-mentioned first filtering operation; when the above-mentioned driving status information of the above-mentioned self-vehicle object does not meet the above-mentioned target measurement point conditions, the first count value matching the above-mentioned inter-frame counter is reduced by 1.

[0014] As an optional scheme, the above-mentioned first acquisition unit is used to: obtain the first point track set collected by the above-mentioned self-vehicle object in the current frame; when the driving state information of the above-mentioned self-vehicle object meets the above-mentioned target measurement point condition, and the number of the above-mentioned first point tracks of the above-mentioned first point track object is greater than the above-mentioned first quantity threshold, perform a first screening operation on the above-mentioned first point track set to obtain the above-mentioned candidate point track set, wherein the number of candidate point tracks of the candidate point track object included in the above-mentioned candidate point track set is less than or equal to the second quantity threshold, and the above-mentioned second quantity threshold is greater than the above-mentioned first quantity threshold.

[0015] As an optional scheme, the above-mentioned first acquisition unit is also used for: when the driving status information of the above-mentioned vehicle object does not meet the above-mentioned target measurement point conditions, and the above-mentioned first point trace quantity of the above-mentioned first point trace object is greater than the above-mentioned first quantity threshold, performing a second screening operation on the above-mentioned first point trace set to obtain a third point trace set, wherein the number of third points of the third point trace object included in the above-mentioned third point trace set is less than or equal to the above-mentioned first quantity threshold; and performing an object recognition operation according to the above-mentioned third point trace set.

[0016] According to another aspect of the embodiments of the present application, a computer program product or a computer program is provided, the computer program product or the computer program including computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the above recognition method based on a point trace set.

[0017] According to another aspect of an embodiment of the present application, there is also provided an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the above-mentioned point trace set-based recognition method through the computer program.

[0018] In the above-mentioned embodiment of the present application, the driving state information of the vehicle object and the candidate point trace set corresponding to the current frame are obtained, wherein the candidate point trace set is obtained through a first screening operation based on the first point trace set collected by the vehicle in the current frame, and the first point trace set includes multiple first point trace objects detected by constant false alarm; when the driving state information of the vehicle object meets the target measurement point condition and the number of first points of the first point trace object is greater than the first quantity threshold, the area boundary description information of the area of ​​interest matching the current frame is obtained; according to the position information corresponding to each of the multiple candidate point trace objects, the multiple candidate point trace objects located in the area of ​​interest are determined as multiple second point trace objects, wherein the number of second points of the multiple second point trace objects is less than or equal to the first quantity threshold; an object recognition operation is performed based on the second point trace set including the multiple second point trace objects, thereby implementing the screening of excess point trace objects based on the area of ​​interest, and performing object recognition based on the screened point trace set, thereby improving the accuracy of object recognition.

[0019] Through the above implementation of the present application, when an excessive number of candidate point trace sets are obtained, combined with the driving status information of the vehicle object, it is dynamically determined whether to use the region of interest to filter the candidate point trace set; when it is determined that the candidate point trace set is filtered according to the region of interest, the region boundary description information of the region of interest of each frame can also be dynamically determined, thereby realizing the dynamic determination of the region of interest, so as to realize the screening of the point trace set based on the dynamically determined region of interest, and solve the technical problems of inaccurate determination of the point trace set and low efficiency in the related technology. Finally, accurate object recognition is realized based on an accurate point trace set, which solves the technical problem of inaccurate object recognition in the existing object recognition method based on the point trace set. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0021] Figure 1 is a schematic diagram of an application environment of an optional recognition method based on a point trace set according to an embodiment of the present application;

[0022] Figure 2 is a flow chart of an optional recognition method based on a point trace set according to an embodiment of the present application;

[0023] Figure 3 is a schematic diagram of an optional recognition method based on a point trace set according to an embodiment of the present application;

[0024] Figure 4is a schematic diagram of another optional recognition method based on a point trace set according to an embodiment of the present application;

[0025] Figure 5 is a schematic diagram of another optional recognition method based on a point trace set according to an embodiment of the present application;

[0026] Figure 6 is a schematic diagram of another optional recognition method based on a point trace set according to an embodiment of the present application;

[0027] Figure 7 is a flowchart of another optional recognition method based on a point trace set according to an embodiment of the present application;

[0028] Figure 8 is a flowchart of another optional recognition method based on a point trace set according to an embodiment of the present application;

[0029] Fig. 9 is a schematic diagram of a recognition device based on a point trace set according to an embodiment of the present application;

[0030] Fig.10 It is a schematic diagram of the structure of an optional electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0032] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0033] According to one aspect of an embodiment of the present application, a recognition method based on a point trace set is provided. Optionally, the recognition method based on a point trace set can be applied to, but is not limited to, Figure 1 In the hardware environment shown. Optionally, the above-mentioned recognition method based on point trace set provided by the present application can be applied to a vehicle terminal. Figure 1 A side view of a vehicle terminal 101 is shown, and the vehicle terminal 101 can travel on a travel surface 113. The vehicle terminal 101 includes an onboard navigation system 103, a memory 102 storing a digitized road map 104, a space monitoring system 117, a vehicle controller 109, a GPS (Global Positioning System) sensor 110, an HMI (Human / Machine Interface) device 111, and also includes an autonomous controller 112 and a telematics controller 114.

[0034] In one embodiment, the space monitoring system 117 includes one or more space sensors and systems, which are used to monitor the visible area 105 in front of the vehicle terminal 101. The space monitoring system 117 also includes a space monitoring controller 118; the space sensors used to monitor the visible area 105 include a laser radar sensor 106, a radar sensor 107, a camera 108, etc. The space monitoring controller 118 can be used to generate data related to the visible area 105 based on the data input from the space sensor. The space monitoring controller 118 can determine the linear range, relative speed and trajectory of the vehicle terminal 101 based on the input from the space sensor, for example, determine the current speed of the vehicle and the relative speed compared to the front vehicle. The space sensor of the vehicle terminal space monitoring system 117 may include an object positioning sensing device, and the object positioning sensing device may include a range sensor, which can be used to locate the front object such as the front vehicle object.

[0035] The camera 108 is advantageously mounted and positioned on the vehicle terminal 101 in a position that allows capturing an image of the visible area 105, wherein at least a portion of the visible area 105 includes a portion of the travel surface 113 in front of the vehicle terminal 101 and including the trajectory of the vehicle terminal 101. The visible area 105 may also include the surrounding environment. Other cameras may also be used, for example, including a second camera disposed on the rear or side portion of the vehicle terminal 101 to monitor the rear of the vehicle terminal 101 and one of the right or left sides of the vehicle terminal 101.

[0036] The autonomous controller 112 is configured to implement autonomous driving or advanced driver assistance system (ADAS) vehicle terminal functionality. Such functionality may include a vehicle terminal onboard control system capable of providing a certain level of driving automation. Driving automation may include a series of dynamic driving and vehicle terminal operations. Driving automation may include a certain level of automatic control or intervention involving a single vehicle terminal function (e.g., steering, acceleration, and / or braking). For example, the above-mentioned autonomous controller can be used to generate a vehicle driving scenario of a vehicle terminal by performing the following steps:

[0037] S102, obtaining driving state information of the vehicle object and a candidate point track set corresponding to the current frame, wherein the candidate point track set is obtained by a first screening operation based on a first point track set collected by the vehicle in the current frame, and the first point track set includes a plurality of first point track objects that pass the constant false alarm detection;

[0038] S104, when the driving state information of the vehicle object meets the target measurement point condition and the number of first point traces of the first point trace object is greater than a first number threshold, obtaining region boundary description information of the region of interest matching the current frame;

[0039] S106, determining the plurality of candidate point trace objects located in the region of interest as a plurality of second point trace objects according to the position information corresponding to each of the plurality of candidate point trace objects in the candidate point trace set, wherein the number of second point traces of the plurality of second point trace objects is less than or equal to a first number threshold;

[0040] S108: performing an object recognition operation according to a second point trace set including a plurality of second point trace objects.

[0041] The HMI device 111 provides human-machine interaction for the purpose of guiding the operation of the infotainment system, GPS (Global Positioning System) sensor 110, onboard navigation system 103 and the like, and includes a controller. The HMI device 111 monitors operator requests and provides the operator with status, service and maintenance information of the vehicle terminal system. The HMI device 111 communicates with multiple operator interface devices and / or controls the operation of multiple operator interface devices. The HMI device 111 may also communicate with one or more devices that monitor biometric data associated with the vehicle terminal operator. For simplicity of description, the HMI device 111 is depicted as a single device, but in the embodiments of the system described herein, it may be configured as multiple controllers and associated sensing devices.

[0042] Operator controls may be included in the passenger compartment of the vehicle terminal 101, and may include, by way of non-limiting example, a steering wheel, an accelerator pedal, a brake pedal, and an operator input device, which is an element of the HMI device 111. The operator controls enable a vehicle terminal operator to interact with the operating vehicle terminal 101 and direct the operation of the vehicle terminal 101 to provide passenger transportation.

[0043] The onboard navigation system 103 uses the digitized road map 104 for the purpose of providing navigation support and information to the vehicle terminal operator. The autonomous controller 112 uses the digitized road map 104 for the purpose of controlling the autonomous vehicle terminal operation or ADAS vehicle terminal functions.

[0044] The vehicle terminal 101 may include a telematics controller 114, which includes a wireless telematics communication system capable of communicating outside the vehicle terminal (including communicating with a communication network 115 having wireless and wired communication capabilities). The wireless telematics communication system includes an off-board server 116 capable of short-range wireless communication with a mobile terminal.

[0045] As an optional implementation, Figure 2 As shown, the recognition method based on the point trace set can be executed by an electronic device, such as Figure 1 In an optional manner, the present embodiment may be applied to an autonomous controller or a radar sensor; in another optional manner, the present embodiment may be applied to a vehicle terminal, which is provided with a radar sensor for transmitting and receiving radar signals, and may further process relevant radar data based on a data processing device provided in the vehicle terminal.

[0046] like Figure 2 As shown, the specific steps of this embodiment include:

[0047] S202, obtaining driving state information of the vehicle object and a candidate point track set corresponding to the current frame, wherein the candidate point track set is obtained by a first screening operation based on a first point track set collected by the vehicle in the current frame, and the first point track set includes a plurality of first point track objects that pass the constant false alarm detection;

[0048] S204, when the driving state information of the vehicle object meets the target measurement point condition and the number of first point traces of the first point trace object is greater than a first number threshold, obtaining region boundary description information of the region of interest matching the current frame;

[0049] S206, determining the plurality of candidate point trace objects located in the region of interest as a plurality of second point trace objects according to the position information corresponding to each of the plurality of candidate point trace objects in the candidate point trace set, wherein the number of second point traces of the plurality of second point trace objects is less than or equal to the first number threshold;

[0050] S208: performing an object recognition operation according to a second point trace set including a plurality of second point trace objects.

[0051] It should be noted that the candidate point track set in the above step S202 can be obtained by screening the first point track set, and the above-mentioned first point track set can be a set of all point track objects monitored according to the radar echo signal, and the above-mentioned first point track object can be a distance-speed RV diagram generated according to the radar echo signal, performing constant false alarm detection (CFAR), and filtering out a part of the noise to obtain the detection point.

[0052] Furthermore, the above-mentioned first screening operation may include but is not limited to being implemented by a point trace filtering algorithm.

[0053] For example, the first screening operation of the first point trace set can be implemented according to the point trace quality of the first point trace object. Specifically, the support vector machine (SVM) algorithm can be used to analyze the features of the first point trace object in the first point trace set, and the distance, pitch angle, signal-to-noise ratio, range spread, azimuth spread, number of echoes, and amplitude information of the first point trace object can be extracted as multi-dimensional feature parameters of the classifier; then, the trained SVM point trace classifier is used to classify the first point trace object according to the extracted multi-dimensional feature parameters to obtain the point trace score of each point trace, and the first point trace set can be screened to obtain the candidate point trace set through the point trace scores corresponding to each of the first point trace objects;

[0054] For another example, the first screening operation of the first point trace set can be implemented according to the risk coefficient of the first point trace object. Specifically, the risk coefficient (between 0 and 1) of the first point trace object can be calculated using the different feature parameter information of the first point trace object, and then some false points can be removed from the first point trace set based on this. The basic method can be to combine the features of the extracted target point traces in pairs, and then calculate the likelihood probability under the feature combination, and finally weighted sum the likelihood probabilities under all feature combinations to obtain the risk coefficient for whether to screen the first point trace object;

[0055] For another example, the distance parameters and speed parameters corresponding to the first point trace objects can be converted into polar coordinates and rectangular coordinates respectively, and the time, speed, azimuth and other information of the point trace are received in each radar processing cycle (for example, one frame processing cycle), the point trace score of the first point trace object is calculated, and the first point trace object is screened according to the point trace score.

[0056] It should be further explained that, in the implementation mode of the present application, the driving state information in the above step S202 may include but is not limited to the first state information for indicating the current driving state of the above-mentioned vehicle object, for example, the vehicle object in the current frame may be in a straight line, a turn, an acceleration, etc. It can be understood that the types of road objects concerned by the vehicle object in different driving states are different, and the distances between the road objects concerned and the current position are also different. Therefore, it can be determined whether to trigger the point trace screening operation based on the area of ​​interest based on the current driving state of the vehicle object;

[0057] In addition, the driving state information may also include but is not limited to the information on the processing method of the vehicle object on the point trace set in the historical frame. It can be understood that the processing method of the vehicle object on the point trace set in the historical frame can be used to indicate the road object information that the vehicle object is concerned about before the current moment, and then determine whether to trigger the point trace screening operation based on the area of ​​interest in the current frame based on the reference information corresponding to the historical frame;

[0058] For example, the above driving status information includes but is not limited to the full point status of the detection points collected in the previous frame. It is understandable that the radar system will set a maximum number of output points per frame when designing, which is usually to balance the real-time processing speed, memory usage and signal processing complexity. Specifically, in some complex or special scenarios, such as when a vehicle passes through an area with extremely high reflectivity (such as a metal fence or tunnel wall), the radar will receive an abnormally large number of echo signals, resulting in a surge in the number of detected points, thereby triggering an "abnormal full point status". When it is detected that the radar signal detected by the vehicle in the historical processing cycle is in an "abnormal full point status", the region of interest screening method can be used synchronously in the current frame to control the number of point objects included in the point set.

[0059] In an optional manner, the above-mentioned obtaining of the driving state information of the vehicle object includes at least one of the following:

[0060] Method 1: Obtaining the road shape information where the vehicle object is located;

[0061] Method 2: obtaining a point trace processing identifier that matches the historical point trace set collected in the previous frame, wherein the point trace processing identifier is used to indicate a processing method for the first point trace set in the current frame;

[0062] Method three: obtain the number of historical point traces of the historical point trace objects included in the historical point trace set.

[0063] In the above-mentioned method 1, the above-mentioned road shape information is used to indicate whether the road of the self-vehicle object in the current frame is a straight line, and then determine whether the self-vehicle object is in a straight driving state. By judging the driving state of the vehicle, further, for example, in a straight driving scene, since dynamic targets may pose a direct threat to driving safety, in order to ensure that key targets are not missed, the radar system needs to adopt a point release strategy to give priority to the point trace output of dynamic or long-distance targets, that is, it is necessary to further perform a screening operation on static or close-range target point traces.

[0064] In the above-mentioned method 2, the point trace processing status of the previous frame can be determined according to the value of the point trace processing identifier flag. For example, in the number of detection points outside the boundary of the area of ​​interest in the historical point trace set collected in the previous frame, when the ratio of the number of uninteresting points to the total number of point traces is greater than the preset threshold, the value of the above point trace processing identifier is determined to be 1, that is, flag = 1, that is, it is determined that the radar data of the previous frame is in an abnormal full point state, and then according to the abnormal identification state of the point trace processing identifier, it can be determined that the point traces that need to be output for object recognition in the current frame can be released, and more detection points that meet the requirements can be released, and then the points of no interest can be further deleted when screening the area of ​​interest, so as to improve the accuracy of the determined point trace set.

[0065] In the above-mentioned method three, the number of historical point traces of the above-mentioned historical point trace object can be the number of point traces in the uninterested area in the Cartesian coordinate system in the point trace output of the previous frame, or the number of points marked as uninterested. It should be noted that when the radar detects the points and converts them into the Cartesian coordinate system, the dynamic target and the static target can be distinguished based on the speed information of the point trace (i.e., the V coordinate). The points of uninterest in the above-mentioned Cartesian coordinate system are specifically static points outside the area of ​​interest. The points marked as uninterested can be specifically determined based on the relationship between the distance unit RIdx corresponding to the boundary point azimIdx corresponding to the boundary point trace output of multiple frames, and then the points outside the area of ​​interest. When the ratio of the uninterested points determined according to the above method to the total number of output points is greater than the preset threshold, it can be judged that the frame belongs to the abnormal full point state, and the marked state value can be used in the next frame to determine whether to trigger the screening operation.

[0066] In addition, in this embodiment, the use of method one, method two and method three can be arbitrarily combined. For example, taking the combination of method two and method three as an example, a point trace processing identifier that matches the historical point trace set collected in the previous frame is obtained, and the point trace processing identifier is in a normal state. The probability of performing special processing on the point release threshold of the current frame is 0.3, and the weight value corresponding to this condition is R1; the number of historical point traces of the historical point trace objects included in the historical point trace set is obtained, for example, the ratio of the number of uninteresting points in the previous frame to the total number of output points is greater than a preset threshold, and the probability of triggering the adjustment operation of the point release threshold of the current frame is 0.8, and the weight value corresponding to this condition is R2. The judgment results of the two judgment conditions are weighted averaged to obtain the result W. When W is greater than the preset threshold, it can be determined to perform a screening operation. Otherwise, no special processing is required.

[0067] Through the above implementation of the present application, the driving status information including road shape information, the point trace processing identifier matched by the historical point trace set collected in the previous frame, and the number of historical point traces of the historical point trace object are obtained, and the trigger conditions for entering the screening operation of the area of ​​interest are designed from different angles. By considering the current road conditions, historical point trace processing strategy and the distribution trend of target point traces, it is ensured that the radar can intelligently adjust its detection and processing process to maintain the best performance and functional stability in various driving environments. Especially in complex and rapidly changing traffic scenes, the execution accuracy of the screening operation is improved, and entering the screening operation under specific conditions improves the efficiency of determining the point trace set.

[0068] Furthermore, in the above step S204, it can be determined whether to trigger a screening operation based on the region of interest according to whether the driving state information meets the target measurement point condition and the number of first point traces in the first point trace set.

[0069] It should be noted that, in this embodiment, since the driving state information in the above embodiment may include a variety of optional information contents, correspondingly, the above target measuring point conditions may match the specific information contents included in the driving state information.

[0070] In addition, in the present embodiment, the above-mentioned condition for determining whether to trigger the screening operation based on the region of interest further includes the comparison result between the first point trace number of the first point trace set and the first quantity threshold. In the present embodiment, the first quantity threshold may be a radar point number threshold determined based on the radar data processing cycle. It is understandable that if the number of points of multiple first point trace objects detected by constant false alarm exceeds the first quantity threshold, and no processing is performed on the first point trace objects, it may cause too many points to be used to perform object detection and further lead to the length of the radar data processing cycle. In the embodiment of the present application, if the first point trace number of the first point trace set is greater than or equal to the first quantity threshold, the first point trace set can be preliminarily screened to obtain a candidate point trace set, and then further based on the candidate point trace set based on the region of interest to obtain a second point trace set for object detection.

[0071] The following further describes the method for determining the target measuring point condition in S204.

[0072] In an optional implementation, after obtaining the driving state information of the vehicle object, the method further includes:

[0073] Case 1: When the road shape information indicates that the vehicle is in a non-straight driving state, it is determined that the driving state information of the vehicle object does not meet the target measurement point conditions;

[0074] Case 2: when the road shape information indicates that the vehicle is in a straight-line driving state and the point trace processing identifier is the first value, it is determined that the driving state information of the vehicle object meets the target measurement point condition;

[0075] Case 3: When the road shape information indicates that the vehicle is in a straight-line driving state, and the weighted sum of the first point trace number and the historical point trace number is greater than the reference quantity threshold, it is determined that the driving state information of the vehicle object meets the target measurement point conditions.

[0076] In the above situation 1, when the road shape information indicates that the vehicle is in a non-straight driving state, it is determined that the driving state information does not meet the target measurement point conditions. For example, when the vehicle turns or changes lanes, the relative position and importance of close-range static targets (such as roadside facilities, walls, etc.) may change. If the screening strategy based on the area of ​​interest is still executed at this time, the point objects that need to be paid attention to may be screened as unimportant point objects, thereby increasing the risk of false alarms and affecting the performance and reliability of the driving assistance system.

[0077] In the above situation 2, when the road shape information indicates that the vehicle is in a straight-line driving state and the point trace processing flag is the first value, it is determined that the driving state information of the vehicle object meets the target measurement point condition.

[0078] For example, when the vehicle is traveling in a straight line, and the ratio of the number of points outside the region of interest in the detection points output in the previous frame to the total number of output points is greater than a certain ratio, resulting in the point processing identifier of the previous frame being the first value (the first value corresponds to an abnormal full point state), it is determined that the target measurement point condition is met. It should be noted that, when the number of first points of the first point trace set initially output in the current frame is greater than the first quantity threshold N_Raw1, combined with the full point state of the previous frame, it can be further determined according to the target measurement point condition whether to perform a point release operation in the process of obtaining a candidate click set based on the first point trace set (i.e., increase the screening threshold of the point trace set to the second quantity threshold N_max1), and when the point release operation is performed, combined with the screening operation of the region of interest, the second point trace set used for the object recognition operation can be further accurately determined.

[0079] Optionally, in the above situation three, when the road shape information indicates that the vehicle is in a straight-line driving state, and the weighted sum of the first point trace number and the historical point trace number is greater than the reference number threshold, it is determined that the driving state information of the vehicle object meets the target measurement point condition. The above historical point trace number can be the first point trace number N_R0 of the first point trace object included in the first point trace set detected by CFAR in the previous frame.

[0080] As an optional implementation, when the vehicle object is in a straight-line driving state, and the weighted sum of the first point trace quantity N_R0 of the first point trace object detected by CFAR in the current frame and the first point trace quantity N_R1 of the first point trace object detected by CFAR in the previous frame is greater than the reference quantity threshold N_Raw2, the temporary point release mechanism is triggered to relax the point trace quantity limit to the second quantity threshold N_max1. And when the first point trace set is screened based on the second quantity threshold N_max1 to obtain a candidate point trace set, the screening operation of the candidate click set based on the region of interest is triggered.

[0081] Through the above implementation methods recorded in this application, the target measurement point conditions are further clarified. If the vehicle is not in a straight-line driving state, the algorithm will not trigger the point release mechanism, avoiding unnecessary relaxation of the point trace number limit when there are a large number of point traces in the close-range area, thereby maintaining the radar's sensitivity and response speed to changes in the close-range environment; using the abnormal full point state detected in the previous frame as the target measurement point condition helps to ensure continuous monitoring and tracking of medium and long-range targets in complex environments. The judgment method based on multi-frame data can filter out erroneous points caused by short-term environmental changes (such as temporary occlusion), ensuring that the point trace number limit is relaxed only when an abnormally high number of point traces is detected in multiple consecutive frames. Through the setting of the above dynamic conditions, the accurate determination of the candidate point trace set entering the area of ​​interest screening operation is achieved.

[0082] The region of interest in the above step S204 is further described below.

[0083] In an optional implementation manner, the acquiring of the region boundary description information of the region of interest matching the current frame includes at least one of the following:

[0084] Method 1: when the driving environment of the ego-vehicle object includes a target enclosure object, the region boundary description information is determined according to the position corresponding to the target enclosure object;

[0085] Method 2: when the driving environment of the vehicle object does not include the target enclosure object, the region boundary description information of the region of interest matching the current frame is determined according to the historical boundary description information corresponding to at least one historical frame.

[0086] like Figure 3 As shown, in the above-mentioned method 1, when the driving environment of the self-vehicle object includes the target enclosure object 302, the boundary information of the region of interest is determined according to the position corresponding to the target enclosure object 302. In an optional manner, the above-mentioned target enclosure object can be a large blocking object such as a fence, a sheet metal wall, etc. It can be understood that when the above-mentioned target enclosure object appears in the driving environment of the self-vehicle object, the road objects that the self-vehicle object needs to pay attention to during driving are usually located in the area related to the above-mentioned enclosure area. For example, when the target enclosure object is a fence object, the road objects that need to be paid attention to are usually located in the region of interest 304 surrounded by the above-mentioned fence object.

[0087] In the above-mentioned method 2, when the driving environment in which the vehicle object is located does not include the target enclosure object, the regional boundary description information of the region of interest matching the current frame can be determined based on the historical boundary description information corresponding to at least one historical frame. It is understandable that since the actual road conditions may be more complicated, for example, when driving in an environment where there are road boundary lines, the road boundary lines can be identified as enclosure objects, thereby increasing the attention paid to the road objects at the road boundary points. However, in actual road conditions, the road boundary lines may suddenly disappear, and then the boundary of interest corresponding to the current frame can be determined based on the historical boundary description information corresponding to the historical frames.

[0088] As an optional implementation, for each historical frame, the system can determine the boundaries of the corresponding historical regions of interest through static point analysis and grid division, record these boundary information, analyze the distribution trend of the boundaries of these historical regions of interest, and identify the law of boundary changes; for example, according to road conditions such as urban roads, highway environments, rural or wild road environments, etc. Then, the division of the region of interest can be determined according to the road classification of the current frame.

[0089] Taking urban roads as an example, the radar system records multiple historical reference frames of urban roads, which contain a large number of dynamic points (vehicles, pedestrians) and static points (buildings, street lights); based on the point data output by the historical reference frame, the system identifies the boundary positions of static obstacles (such as buildings, street lights, and curbstones) that may exist on both sides of the urban roads, and uses them as the boundaries of the area of ​​interest. On urban roads, the area at a certain distance ahead is usually focused on to ensure that vehicles and pedestrians ahead are detected in time. The radar system sets the area of ​​interest for urban roads to 100-200 meters directly in front of the vehicle, with a width covering all lanes. Therefore, when the current frame is an urban road, the setting method of the area of ​​interest corresponding to the historical frame of the above-mentioned urban road can be obtained to determine the area of ​​interest that matches the current frame.

[0090] In another optional manner, the historical boundary description information corresponding to each of the multiple frames of radar data before the current frame can be obtained to determine the boundary of interest corresponding to the current frame. Preferably, the attention weights corresponding to the multiple historical boundary description information can be configured, for example, the first weight can be configured for the 3 frames of historical boundary description information closest to the current frame, the second weight can be configured for the 4-6 frames of historical boundary description information closest to the current frame, and the third weight can be configured for the 6-10 frames of historical boundary description information closest to the current frame, and then the boundary of interest corresponding to the current frame can be determined according to the weighted sum of the 10 frames of historical boundary description information.

[0091] Optionally, in the third method, when the driving environment of the vehicle object does not include the target enclosure object, the preset boundary information of the region of interest can also be obtained. Optionally, the region of interest can be preset as a medium and long distance range in front of the vehicle according to vehicle safety and assisted driving requirements, and the boundary information of the region of interest can be preset to be between 50 meters and 200 meters, and the azimuth covers 0° to 120° and 240° to 360°. This is only an example.

[0092] In an optional manner, the above-mentioned boundary information of the area of ​​interest can also be the maximum value of the horizontal coordinate of the area of ​​interest. Specifically, the boundary can be identified based on the distribution of stationary points using a grid division method. First, based on the Cartesian coordinate system of the vehicle, the stationary points are grouped by grid in terms of the horizontal position, and the two groups with the largest number of points in the horizontal grid are recorded. Then, the distribution of these two groups in the vertical position is counted. If the density of the point distribution in the vertical position meets certain conditions, the maximum value of the horizontal position of the two groups of points is recorded as the horizontal boundary value of the area of ​​interest.

[0093] Through the above implementation described in the present application, the scope of the area of ​​interest is dynamically determined according to the current road conditions, which improves the accuracy of screening out points of no interest, thereby improving the accuracy of the point set for object recognition operations.

[0094] The following further describes the screening operation based on the region of interest in S206.

[0095] In an optional implementation manner, the above-mentioned determining the multiple candidate point trace objects located in the region of interest as multiple second point trace objects includes at least one of the following:

[0096] Method 1: when the region boundary description information includes the plane coordinates of at least one region boundary point, a plurality of second point trace objects are determined from a plurality of candidate point trace objects according to a comparison result of the plane coordinates of the candidate point trace object and the plane coordinates of at least one region boundary point;

[0097] Method 2: When the area boundary description information includes a boundary point description array corresponding to at least one area boundary point, multiple second point trace objects are determined from multiple candidate point trace objects based on the comparison result between the azimuth index identifier of the candidate point trace object and the reference distance unit index identifier included in the boundary point description array, wherein the boundary point description array is used to indicate the correspondence between the boundary distance and the azimuth index of the boundary point.

[0098] It can be understood that in the above-mentioned method 1, when the area boundary description information includes the plane coordinates (x, y) of at least one area boundary point, multiple second point objects can be determined from multiple candidate point objects based on the comparison results of the plane coordinates of the candidate point objects with the plane coordinates of at least one area boundary point.

[0099] Preferably, the above-mentioned method 2 can be used to perform the screening operation based on the region of interest.

[0100] Before performing the region of interest screening operation through the above-mentioned method 2, multiple candidate point trace objects in the candidate point trace set can be input into the first angle measurement module to obtain the azimuth index identifiers corresponding to the multiple candidate point trace objects; the above-mentioned first angle measurement module can specifically be an FFT angle measurement module, which is used to output the azimuth index number azimIdx corresponding to the point trace.

[0101] Furthermore, in the above-mentioned method 2, the second point set can be subjected to a key screening operation according to the boundary information of the region of interest and the azimuth index identifiers corresponding to each of the plurality of second point objects. The above-mentioned boundary information of the region of interest includes but is not limited to the azimuth index identifiers corresponding to the boundary points and the distance units (RIdx) matching the boundary points.

[0102] It should be noted that linear frequency modulation: R = RIdx*Rres; V = VIdx*Vres; step frequency: R = RIdx*Rres-fc*VIdx*Vres / k; V = VIdx*Vres-Δf*RIdx*Rres / fc / PRT; Rres is the distance resolution, Vres is the velocity resolution, fc is the radar center frequency, Δf is the frequency step between adjacent chirp signals, k is the frequency modulation slope, and PRT is the chirp signal pulse repetition period.

[0103] And, the lateral position: xpos = R*cos(θ)+longoffset, the longitudinal position: ypos = R*sin(θ)+latoffset, where θ = azimuth0+yawAngle0, azimuth0 is the point azimuth calculated by the azimuth index azimIdx0, R is the target distance, longoffset is half the length of the vehicle, latoffset is half the width of the vehicle, and yawAngle0 is the radar installation angle used for boundary statistics. That is, based on the output azimuth index number azimIdx and the distance unit (RIdx), the position coordinates of the detection point can be determined. Based on the azimIdx-RIdx of the boundary point, the second point set can be screened.

[0104] Through the above implementation methods recorded in this application, multiple candidate point trace objects in the candidate point trace set are input into the first angle measurement module to obtain the azimuth index identifiers corresponding to each of the multiple candidate point trace objects; obtain the boundary information of the region of interest that matches the current frame; and perform key screening operations on the candidate point trace set based on the boundary information of the region of interest and the azimuth index identifiers corresponding to each of the multiple second point trace objects. In a complex driving environment, such as a road section with a large number of barriers or reflectors, the radar may receive many points from short-range areas of no interest. These points may occupy the radar's memory space, resulting in the inability to output target points in medium and long-range areas of interest normally. Through ROI boundary screening, it can be ensured that even when this happens, medium and long-range targets of interest are given priority, avoiding false alarms or omissions caused by full memory points, and improving the accuracy of point trace set determination.

[0105] In an optional implementation, the above-mentioned determination of multiple second point trace objects from multiple candidate point trace objects based on the comparison result between the azimuth index identifier of the candidate point trace object and the reference distance unit index identifier included in the boundary point description array includes:

[0106] S1, traversing multiple candidate point trace objects, and determining a reference distance unit index identifier matching the azimuth index identifier from a boundary point description array according to the azimuth index identifier of the currently traversed candidate point trace object;

[0107] S2, determining a reference azimuth index identifier corresponding to a reference distance unit index identifier according to a boundary point description array;

[0108] S3: When the azimuth index identifier of the candidate point trace object matches the reference azimuth index identifier, the currently traversed candidate point trace object is determined as the second point trace object.

[0109] In the above implementation, the boundary information of the region of interest is first determined, wherein the boundary point description array is used to indicate the association relationship between the azimuth index identifier and the distance unit index identifier.

[0110] Specifically, each detection point input FFT angle measurement module will obtain the corresponding azimuth index azimIdx, and after the echo signal is processed by Fourier transform of distance and Doppler dimension, the distance unit index identifier RIdx and VIdx of the detection point can be obtained. The azimIdx-RIdx relationship of the regional boundary points of the region of interest can be counted to limit the marking boundary; in addition, when the above-mentioned offset angle (that is, the radar installation angle yawAngle changes, it will affect θ, causing the boundary of the region of interest written in the Cartesian coordinate system to shift), the corresponding azimIdx-RIdx boundary will also shift, so the deviation caused by the installation angle needs to be compensated before the boundary value comparison is performed. The specific compensation formula is as follows: azimIdx = azimIdx0 + (yawAngle-yawAngle0) / azi_step; where yawAngle is the radar installation angle after the change, and azi_step is the angle measurement step (unit °). According to the above steps, the boundary information of the region of interest can be corrected. Specifically, the one-dimensional array gARIdx (i.e., the boundary point description array) is corrected. The one-dimensional array gARIdx (i.e., the boundary point description array) represents the relationship between the angle index and the distance index of the boundary points of the region of interest.

[0111] Further, determine the reference distance unit index identifier that matches the azimuth index identifier of the second point object in the boundary information of the area of ​​interest, and the reference azimuth index identifier that matches the reference distance unit index identifier; when the azimuth index identifier of the second point object does not match the reference azimuth index identifier, determine to delete the second point object.

[0112] As an optional implementation, for example, the maximum distance unit RmaxIdx of the radar is 200, the maximum azimuth index AmaxIdx is 360 (corresponding to 360° omnidirectional coverage), and the region of interest is from 50 to 200 distance units in distance, and from 0° to 120° and 240° to 360° in azimuth (i.e., the left and right sides in front of the radar). The boundary information of this region of interest is represented by a one-dimensional array gARIdx, such as: the first half of gARIdx (from index 50 to index 169) stores the boundary values ​​of azimuth 0° to 120°, that is, the left boundary value of each distance unit is azimIdx=0, and the right boundary value is azimIdx=120. The second half of gARIdx (starting from index 201 to index 319) stores the boundary values ​​of the azimuth angle from 240° to 360°, that is, the left boundary value of each distance unit is azimIdx=240, and the right boundary value is azimIdx=360.

[0113] Assume that there is a second point object with a (RIdx, azimIdx) value of (100, 80), and find the azimuth boundary value corresponding to RIdx=100 in the gARIdx array. For example, gARIdx

[100] is the left boundary value azimIdx=0, and gARIdx[100+1] (i.e., gARIdx

[101] ) is the right boundary value azimIdx=120. It should be noted that, considering the continuity and storage efficiency of the gARIdx array, the left boundary value and the right boundary value are usually stored adjacent to each other.

[0114] According to the above steps, if the azimIdx value 80 of the second trace object is within the left boundary value 0 and the right boundary value 120, the second trace object is determined to be within the region of interest, otherwise the second trace object outside the region of interest needs to be deleted.

[0115] Through the above implementation methods recorded in this application, the boundaries of the region of interest are dynamically adjusted according to different driving scenarios, so that high target detection performance can be maintained in various driving environments, and the adaptability and flexibility of the radar system are enhanced. In addition, the ROI boundary screening not only limits the range of radar processing points, but also improves the accuracy of point trace azimuth measurement by compensating for the influence of the radar installation angle. This means that the system can more accurately identify which points are located in the region of interest, so as to perform more in-depth processing and analysis on these points, and improve the detection accuracy of actual targets.

[0116] In an optional implementation, after determining the plurality of candidate point trace objects located in the region of interest as the plurality of second point trace objects according to the position information corresponding to each of the plurality of candidate point trace objects, the method further includes at least one of the following:

[0117] Case 1: when the ratio between the first reference point trace quantity of the reference point trace object and the second point trace quantity of the second point trace object is greater than the first threshold, the point trace processing identifier of the current frame is determined to be the first value, wherein the reference point trace object is a point trace object outside the boundary of the region of interest;

[0118] Case 2: when the ratio between the second reference point quantity of the reference point object and the second point quantity of the second point object is greater than the second threshold, the point processing identifier of the current frame is determined to be the first value.

[0119] Optionally, in method 1, the first reference point trace number of the reference point trace object can be the number of static point traces outside the boundary of the area of ​​interest obtained by calculating the horizontal and vertical coordinate information (xypos) of the point trace in the Cartesian coordinate system. It should be noted that the Cartesian coordinate system calculation can distinguish between dynamic and static point traces. In this method, the ratio of the number of static point traces of no interest outside the area of ​​interest to the number of point traces entering the angle measurement module is calculated, and when the ratio result is greater than the first threshold, it is determined to be in an abnormal full point state. The above first value can be 1, that is, the point trace processing flag of the current frame is set to 1, which can be used to indicate that the radar data corresponding to the current frame is in an abnormal full point state.

[0120] In the above-mentioned method 2, the number of second reference points of the reference point object can be the number of points outside all the areas of interest determined by generating the azimIdx-RIdx boundary of interest through multiple historical frames, and the ratio of the number of points marked as not interested outside the area of ​​interest to the number of points entering the angle measurement module is calculated. When the ratio result is greater than the second threshold, it is determined to be in an abnormal full point state.

[0121] That is, if the number of points M output by the current frame (D1 frame) is greater than c1*initial preset output threshold, or the number of points N entering the angle measurement module is greater than or equal to the initial preset output threshold, then the number of points n1 in the uninterested area in the Cartesian coordinate system is counted. If n1 is greater than c2 (i.e., the first threshold)*N, or the number of points marked as uninteresting is greater than c3 (i.e., the second threshold)*N, then the current frame is considered to be in an abnormal full point state, that is, the point processing flag of the current frame is set to 1, and then the click processing information of the current frame is passed to the next frame according to the above point processing flag. Correspondingly, the point processing flag passed by the previous frame obtained by the current frame can be used to indicate the point processing state of the previous frame.

[0122] Through the above implementation methods recorded in this application, the full point state is judged according to the ratio of the number of points of no interest to the number of input points, so that the millimeter wave radar system can effectively monitor the distribution of points in the current frame and promptly detect abnormal full point states. This monitoring mechanism helps the system to quickly adjust its processing strategy when the number of points increases abnormally, avoiding processing delays or performance degradation caused by this. By setting different trigger conditions to adjust the point release restrictions and the screening operation restrictions of the area of ​​interest in different situations, it can ensure that all potential dynamic targets can be detected, and improve the system's adaptability and response speed to complex environments; and then the processing of static or irrelevant points can be reduced by marking the points in the area of ​​interest, reducing the false alarm rate and improving the accuracy of determining the point set.

[0123] In an optional implementation manner, after determining the point trace processing flag of the current frame as the first value, the method further includes:

[0124] S1, modifying the value of the inter-frame counter matching the current frame to a first count value, wherein the first count value is greater than 0, and the value of the inter-frame counter is used to indicate the number of frames for performing a first filtering operation;

[0125] S2: when the driving state information of the vehicle object does not meet the target measurement point condition, the first count value matched with the inter-frame counter is reduced by 1.

[0126] In the above steps S1-S2, the specific value of the first count value can be determined according to the application scenario, and the value NumCnt of the inter-frame counter is set to U. In the next U frame, even if the judgment condition of the full point state is no longer met, the system will automatically relax the limit on the number of points, allowing more points to enter the subsequent processing stage, which is helpful in ensuring that the detection of medium and long-range targets of interest will not be affected by the memory usage of short-range uninteresting points in complex environments (such as roads with a large number of barriers or reflectors).

[0127] Further, when NumCnt is reduced to 0, the system will re-judge the full-point status based on the number of points and the full-point mark flag to decide whether to continue to trigger the point release mechanism or restore the normal point number limit.

[0128] For example, after the first judgment triggers the filtering operation, the inter-frame counter value is set to 3 (assuming it corresponds to the D2 frame), and the filtering operation is directly performed from the D3 frame to the D5 frame. For example, the output point trace restriction is relaxed. If the D6 frame determines that the target measurement point conditions are not met based on the vehicle driving state (for example, the vehicle is in a non-straight driving state), the NumCnt value is reduced by 1 until the filtering operation is exited. The above implementation method reduces the number of times the flag judgment abnormal full point state is entered, thereby improving the efficiency of determining the point trace set.

[0129] In an optional implementation, the above-mentioned obtaining the driving state information of the vehicle object and the candidate point track set corresponding to the current frame includes:

[0130] S1, obtain the first point trace set collected by the vehicle object in the current frame;

[0131] S2, when the driving state information of the vehicle object meets the target measurement point condition and the number of first point traces of the first point trace object is greater than the first quantity threshold, a first screening operation is performed on the first point trace set to obtain a candidate point trace set, wherein the number of candidate point traces of the candidate point trace objects included in the candidate point trace set is less than or equal to the second quantity threshold, and the second quantity threshold is greater than the first quantity threshold.

[0132] Optionally, after obtaining the driving state information of the vehicle object, the method further includes:

[0133] S1, when the driving state information of the vehicle object does not meet the target measurement point condition and the number of first point traces of the first point trace object is greater than the first quantity threshold, performing a second screening operation on the first point trace set to obtain a third point trace set, wherein the number of third point traces of the third point trace object included in the third point trace set is less than or equal to the first quantity threshold;

[0134] S2, performing an object recognition operation according to the third point trace set.

[0135] It can be understood that in this embodiment, when the first point trace set is obtained, the screening threshold condition can be relaxed according to the method described in the above embodiment, and the first screening operation can be performed. And when the candidate point trace set is screened, the second point trace set for performing object recognition is obtained through the above-mentioned screening operation of the region of interest.

[0136] In the case of another branch, that is, the driving state information of the vehicle object does not meet the target measurement point conditions, and the number of first points of the first point object is greater than the first number threshold, a second screening operation is performed on the first point set to obtain a third point set, and an object recognition operation is performed based on the third point set.

[0137] It can be understood that since the first quantity threshold N_max0 corresponding to the second filtering operation is smaller than the second quantity threshold N_max1 corresponding to the first filtering operation, that is to say, the above-mentioned candidate point set is obtained through a filtering operation that temporarily enlarges the point quantity threshold, and then combined with the subsequent filtering operation of the area of ​​interest, the point set can be accurately obtained.

[0138] When the driving state information of the vehicle object does not meet the target measurement point condition and the number of first points of the first point object is greater than the first number threshold, the first point set is filtered according to the preset first number threshold N_max0 to obtain the third point set, thereby skipping the filtering operation of the area of ​​interest.

[0139] Through the above-mentioned implementation mode of the present application, when the driving state information of the self-vehicle object meets the target measurement point condition and the number of first point traces of the first point trace object is greater than the first quantity threshold, the first screening operation is performed on the first point trace set to obtain a candidate point trace set, and the output point trace threshold limit is adjusted and amplified according to the self-vehicle state and the number of collected detection points, and the long-distance interesting points are released; in addition, a key screening operation is performed on the candidate point trace set to obtain a second point trace set, and the points of no interest are screened and deleted, thereby improving the efficiency and accuracy of object recognition; and, when the driving state information of the self-vehicle object does not meet the target measurement point condition and the number of first point traces of the first point trace object is greater than the first quantity threshold, the first point trace set is subjected to a second screening operation to obtain a third point trace set, and the output point trace quantity threshold is limited in consideration of the memory space limitation and the processing capacity, thereby improving the processing efficiency of the points, and solving the technical problem that the point trace set cannot be accurately and efficiently determined in the prior art.

[0140] The following combination Figure 7 Further explanation of a method combining a temporary placement scheme and an area of ​​interest screening method

[0141] S702, initializing the inter-frame counter NumCnt=0;

[0142] S704, echo signal acquisition and CFAR detection, outputting the range Doppler units of all detection points;

[0143] In the above step S704, the echo signal of the current frame (D1 frame) can be obtained. The radar echo signal can be generated by either general linear frequency modulation or step frequency modulation continuous wave radar, that is, the radar transmits electromagnetic waves to the target of interest, thereby receiving the echo signal reflected from the target of interest.

[0144] Next, the echo signal of the current frame (D1 frame) is processed to obtain a range-velocity map. The range-velocity map (RV map) of the echo signal is obtained by performing Fourier transform processing of the range and Doppler dimension on the echo signal, which contains the range-velocity information of all target points and clutter points in the current environment.

[0145] Next, based on the range-velocity graph (RV graph), the number of points detected by CFAR in the current frame (D1 frame) and the corresponding range-Doppler unit index are output. In this step, according to the range-velocity RV graph generated by the radar echo signal, the constant false alarm detection (CFAR) is normally performed, and the points that can pass the CFAR detection are determined as the first point object, and the points that do not pass the CFAR are the noise points. However, due to the complexity of the road environment, even the points that pass the CFAR detection contain noise points that are not of interest. Here, the first point number N_R1 of the first point object detected by CFAR and the corresponding range unit index RIdx and DopIdx are preliminarily output.

[0146] S706, combining the previous frame flag and the current frame information to determine whether to trigger the temporary placement determination module, if not triggered, set the inter-frame counter NumCnt-1 (when NumCnt>0);

[0147] In the above step S706, the previous frame flag and the current frame information are the specific information content included in the above driving status information, wherein the previous frame flag is a specific point trace processing identifier, and the current frame information includes the current road information.

[0148] The specific process of determining the temporary placement point is described below.

[0149] In the process of judging whether it is necessary to perform a temporary point release operation, the flag_0 of the previous frame (D0 frame) and the information of the current frame (D1 frame) are combined to determine whether to trigger the temporary point release mechanism and confirm whether to relax the limit on the number of points entering the channel matching, distance and speed calculation modules.

[0150] The specific judgment logic is as follows: when the vehicle is in a straight-moving state, it enters the temporary point release judgment module, otherwise the temporary point release mechanism is not triggered, the point trace number limit also maintains the original value N_max0 (ie, the first quantity threshold), and the inter-frame counter NumCnt is reduced by 1 (ie: U-1).

[0151] In the process of performing temporary point placement judgment, if condition 1) the number of points detected by CFAR in the current frame NR1 is greater than N_Raw1 and the previous frame flag = 1, or condition 2) the weighted sum of the number of points detected by CFAR in the current frame (N_R0, the first number of points) and the number of points detected by CFAR in the previous frame (N_R1) is greater than N_Raw2, the temporary point placement mechanism is triggered, that is, the limit on the number of points is relaxed to N_max1 (that is, the second number threshold), otherwise, the limit on the number of points still maintains the original value Nmax0 (that is, the first number threshold), and the inter-frame counter NumCnt is reduced by 1 (that is: U-1).

[0152] Among them, N_Raw1 and N_Raw2 are the number of CFAR detection points statistically based on the abnormal full-point state. They are positive integer fixed values that can be dynamically adjusted according to the actual situation. Generally, N_Raw2 > N_Raw1. They can be the results of multiple frames of Nraw1 and Nraw2 obtained by comparing the full-point and non-full-point states through a histogram on the matlab side. (Generally, the result in the non-full-point state is less than that in the full-point state). Take the minimum value and median in the full-point state as Nraw1 and Nraw2 respectively. flag_0 (i.e., the point track processing identifier) is a flag bit representing whether the data in the previous frame is in the abnormal full-point state, and the calculation method of the current frame flag_1 in the subsequent steps is the same.

[0153] It can be understood that the above-mentioned third point track set in the above-mentioned embodiment can be obtained through the second screening operation performed based on the original threshold Nmax0 (i.e., the first quantity threshold); the candidate point track set in the above-mentioned embodiment can be obtained through the first screening operation performed based on the temporary threshold N_max1 (i.e., the second quantity threshold).

[0154] S708, calculate the distance and speed of all detection points;

[0155] Next, calculate the distance and speed information of the candidate point track objects in the N candidate point track sets output after the above screening operation. According to the distance unit RIdx and Doppler unit VIdx information, calculate and output the distance R and speed V information of all candidates.

[0156] Specifically, for linear frequency modulation: R = RIdx * Rres; V = VIdx * Vres; for stepped frequency: R = RIdx * Rres - fc * VIdx * Vres / k; V = VIdx * Vres - Δf * RIdx * Rres / fc / PRT; Rres is the distance resolution, Vres is the speed resolution, fc is the radar center frequency, Δf is the frequency step between adjacent chirp signals, k is the frequency modulation slope, and PRT is the chirp signal pulse repetition period.

[0157] In the second screening mode, the number of detection points N entering this module < N_max0, and after successfully triggering the temporary point release mechanism (i.e., the first screening mode), the number of points N entering this module < N_max1, where Nmax0 < Nmax1. In addition, due to the existence of the point track number limit, the number of points output in this step is generally less than or equal to the number of points output by CFAR detection, that is, N <= NR1.

[0158] S710, is the number of detection points > Nmax0?

[0159] If the number of detection points > Nmax0, the point release strategy is entered, and the number of output detection points is greater than the original output threshold Nmax0. Enter the following steps to filter the points of interest through the boundaries of the region of interest;

[0160] S712, channel calibration and FFT angle measurement, output and save the FFT angle measurement results azimIdx of all detection points; otherwise, execute S717, if there are invalid points marked, they should be removed, and the total number of points entering the angle measurement module should be less than Nmax0.

[0161] S714, compensating for the influence of the installation angle azimIdx;

[0162] In the above steps, the compensation of the influence of the installation position and installation angle changes: when the radar installation angle yawAngle changes, it will affect θ, causing the boundary of the area of ​​interest written in the Cartesian coordinate system to shift (such as Figure 5 As shown in the figure, the corresponding azimIdx-RIdx boundary will also be offset, so the deviation caused by the installation angle needs to be compensated before comparing the boundary values. The specific compensation formula is as follows: azimIdx = azimIdx0 + (yawAngle-yawAngle0) / azi_step; where yawAngle is the changed radar installation angle, and azi_step is the angle measurement step (unit: °).

[0163] S716, according to the azimIdx-Ridx relationship of the boundary of interest, mark the points outside the boundary as invalid points;

[0164] In the above steps, since the scenes where the targets in the area of ​​interest are missed are mostly found when the vehicle is driving straight through an environment with strong reflective targets such as large fences, the number of target points in the near lateral position that are farther away exceeds the limit of the number of points Nmax0, resulting in the inability to detect medium and long-distance moving targets. By limiting the horizontal and vertical positions of the output points in the Cartesian coordinate system, the memory occupation of the near-range uninteresting points can be released, which can be used to output the medium and long-distance target points of interest. The calculation of the horizontal and vertical positions (xpos and ypos) involves the calculation of angles and the conversion of the coordinate system, which will increase the additional running time. In addition, it is found from the following formula that after the azimIdx and RIdx of a point are determined, its corresponding xpos and ypos are also determined. Therefore, here, the azimIdx-RIdx relationship of the boundary points is directly counted based on the interested boundaries accumulated in multiple frames to limit the marking boundaries.

[0165] Lateral position: xpos=R*cos(θ)+longoffset, longitudinal position: ypos=R*sin(θ)+latoffset, where θ=azimuth0+yawAngle0, azimuth0 is the point azimuth calculated by the azimuth index azimIdx0, R is the target distance, longoffset is half the vehicle length, latoffset is half the vehicle width, and yawAngle0 is the radar installation angle used for boundary statistics.

[0166] In addition, considering the action areas of special functional scenarios such as RCTA (rear cross traffic alert system) and FCTA (front cross traffic alert system), the points in the close-range designated area are also relaxed, and finally the boundary of the area of ​​interest in the Cartesian coordinate system is generated as follows Figure 4 As shown, the corresponding generated azimIdx-RIdx boundary of interest is as follows Figure 5 The final output of the boundary of interest gARIdx is a one-dimensional array, which represents the relationship between the angle index and the distance index of the boundary points of the region of interest. The size of the array is determined by the selected angle measurement step azi_step.

[0167] S718, if there are invalid points marked, they should be removed, and the total number of points entering the angle measurement module should be less than N_max0 (a first number threshold);

[0168] S720, NumCnt>0? If NumCnt>0, execute S720-1, flag=1; S720-2, flag NumCnt; that is, continue to judge the trigger condition of the next frame according to flag and NumCnt.

[0169] S722, comprehensively consider the invalid points and total points of the current frame, output the global variable flag to participate in the next frame judgment, and if flag = 1, set NumCnt = U;

[0170] In an optional manner, the second number threshold N_max1 involved in the above-mentioned first screening operation process can also be dynamically determined according to the state of the previous frame. Figure 8 A method for dynamically determining the second quantity threshold N_max1 is described.

[0171] Flowchart Figure 8 The above process is explained as follows:

[0172] S802, evaluating the frame duration and memory situation to determine the maximum number of points placed in the system (a first number threshold);

[0173] In the above steps, the frame duration, memory conditions, and radar performance requirements can be evaluated to determine the maximum number of temporary points N_max0 and the maximum number of points M_max for the final output. Set different numbers of points, count the time required for all processing modules of the point traces under different numbers of points, and record the maximum number of points that meet the time requirements of a single frame. At the same time, evaluate the remaining memory space. If a certain number of points occupies more memory than the existing memory space limit, reduce the maximum number of points according to memory usage to obtain N_max0. Combine radar performance and customer needs to jointly determine the maximum number of output points M_max of the radar, where M_max<=N_max0.

[0174] S804, obtaining the echo signal of the current frame and performing constant false alarm detection normally;

[0175] Get the radar echo signal of the current frame and output the target point of the current frame. Perform constant false alarm detection (CFAR) on the echo signal of the current frame, and normally output all target points that pass the CFAR detection, and record the number of points N_R1 (the first point trace number) that pass the CFAR detection in the current frame.

[0176] S806, obtaining and calculating the echo signal processing time of the current frame, and calculating the ratio of the current frame time to the theoretical frame time;

[0177] According to the input clock counter value and CPU operating frequency, the frame duration T0 of the current frame is calculated. The frame duration of the previous frame can be obtained by multiplying the clock counter difference between the current frame and the previous frame by the CPU operating frequency. Since the frame duration is at the millisecond level, it is assumed here that the point trace characteristics in two consecutive frames change little, and the frame duration between the corresponding frames also changes little, that is, the calculated frame duration of the previous frame data can be recorded as the frame duration of the current frame. Next, according to the frame duration of the current frame obtained in S1, its ratio to the theoretical maximum frame duration is calculated.

[0178] S808, determining the number of points to be placed in the current frame (a second number threshold) by combining the frame duration ratio of the current frame and the percentage of points outside the boundary of interest of the previous frame;

[0179] The number of temporary points N_max1 of the current frame is determined based on the ratio of the proportion of uninteresting points in the previous frame to the frame duration of the current frame, as well as the number of points detected by CFAR in the current frame N_R1. Assuming that the number of points is divided into m blocks, the value δ obtained by weighted summation of the point proportion and frame duration ratio is m When in different intervals, the number of points N_max1 of the current frame is also assigned different values ​​accordingly.

[0180] δ m =α1*p before / M before +α2*T now / Tmax , where α1 and α2 are coefficients less than 1 and α1+α2=1, p before / M before Indicates the ratio of the number of uninteresting points in the previous frame to the total number of points output in the previous frame, T now / T max Indicates the ratio of the current frame duration to the theoretical maximum frame duration, and is greater than the maximum number of output points Mmax determined according to radar performance requirements, that is, M_max<=N_max0 <N_max1。

[0181] S810, dynamically determine the boundary of the region of interest of the current frame and count the number of points outside the boundary, and calculate the proportion of uninteresting points in the current frame for participating in the point placement judgment of the next frame.

[0182] Perform distance and speed measurement calculations on all the output N_m points (candidate point trace objects), and finally output M_now points (second point trace objects) to enter the angle measurement module. Perform distance and speed measurement calculations on all the points released in the above steps to obtain the distance and speed information of the target. Before entering the angle measurement module, delete the points outside the boundary of the area of ​​interest according to the point trace features to ensure that only M_now points enter the angle measurement module.

[0183] The boundary of the region of interest is dynamically determined based on the point characteristics of the current frame data, and the number of uninteresting points outside the boundary of interest is counted. It is generally believed that full points are likely to appear in scenes with fences and iron walls, so here the fence / iron wall is considered as the boundary of the region of interest, the boundary position is determined by grid division, and the number of uninteresting points outside the boundary p_now is counted.

[0184] According to the distribution of stationary points, the boundary is identified by grid division. First, based on the Cartesian coordinate system of the vehicle, the stationary points are grouped by grid in the horizontal position, and the two groups with the largest number of points in the horizontal grid are recorded. Then, the distribution of these two groups in the vertical position is counted. If the density of the point distribution in the vertical position meets certain conditions, the maximum value of the horizontal position of the two groups of points is recorded as the horizontal boundary value of the area of ​​interest.

[0185] Finally, the proportion of uninteresting points in the current frame is calculated to participate in the point placement judgment of the next frame. The uninteresting points in the current frame are divided by the total number of output points to obtain the proportion of uninteresting points in the current frame pnow / Mnow.

[0186] A method for dynamically switching the number of millimeter-wave radar measurement points provided by the above-mentioned embodiment of the present application is used to dynamically switch the maximum number of points in different gears in combination with environmental characteristics and radar frame time consumption, so as to minimize the frame time consumption while ensuring that the targets in the area of ​​interest are released.

[0187] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0188] According to another aspect of the embodiment of the present application, a point trace set-based recognition device for implementing the above-mentioned point trace set-based recognition method is also provided. Fig. 9 As shown, the device comprises:

[0189] A first acquisition unit 902 is used to acquire driving state information of the ego vehicle object and a candidate point track set corresponding to the current frame, wherein the candidate point track set is obtained by a first screening operation based on a first point track set collected by the ego vehicle in the current frame, and the first point track set includes a plurality of first point track objects that pass the constant false alarm detection;

[0190] The second acquisition unit 904 is used to acquire the region boundary description information of the region of interest matching the current frame when the driving state information of the vehicle object meets the target measurement point condition and the number of the first point traces of the first point trace object is greater than the first number threshold;

[0191] A determination unit 906 is used to determine the plurality of candidate point trace objects located in the above-mentioned area of ​​interest as a plurality of second point trace objects according to the position information corresponding to each of the plurality of candidate point trace objects in the above-mentioned candidate point trace set, wherein the number of second point traces of the plurality of second point trace objects is less than or equal to the first number threshold;

[0192] The recognition unit 908 is used to perform an object recognition operation according to the second point trace set including a plurality of the second point trace objects.

[0193] As an optional scheme, the above-mentioned determination unit 906 includes at least one of the following: a first determination module, which is used to determine a plurality of second point trace objects from a plurality of the above-mentioned candidate point trace objects based on a comparison result between the plane coordinates of the above-mentioned candidate point trace objects and the plane coordinates of at least one of the above-mentioned area boundary points when the above-mentioned area boundary description information includes the plane coordinates of each of the at least one area boundary points; a second determination module, which is used to determine a plurality of second point trace objects from a plurality of the above-mentioned candidate point trace objects based on a comparison result between the azimuth index identifier of the above-mentioned candidate point trace object and the reference distance unit index identifier included in the above-mentioned boundary point description array when the above-mentioned area boundary description information includes a boundary point description array corresponding to at least one area boundary point, wherein the above-mentioned boundary point description array is used to indicate the correspondence between the boundary distance of the boundary point and the azimuth index.

[0194] As an optional scheme, the above-mentioned second acquisition unit 904 includes at least one of the following: a first acquisition module, which is used to determine the above-mentioned area boundary description information according to the position corresponding to the above-mentioned target enclosure object when the driving environment of the above-mentioned self-vehicle object includes the target enclosure object; a second acquisition module, which is used to determine the area boundary description information of the area of ​​interest matching the above-mentioned current frame according to the historical boundary description information corresponding to at least one historical frame when the driving environment of the above-mentioned self-vehicle object does not include the target enclosure object.

[0195] As an optional scheme, the above-mentioned second determination module includes: traversing the above-mentioned multiple candidate point trace objects, and determining the reference distance unit index identifier that matches the above-mentioned azimuth index identifier from the above-mentioned boundary point description array according to the azimuth index identifier of the above-mentioned candidate point trace object currently traversed; determining the reference azimuth index identifier corresponding to the above-mentioned reference distance unit index identifier according to the above-mentioned boundary point description array; when the above-mentioned azimuth index identifier of the above-mentioned candidate point trace object matches the above-mentioned reference azimuth index identifier, determining the above-mentioned candidate point trace object currently traversed as the above-mentioned second point trace object.

[0196] As an optional solution, the first acquisition unit 902 is used for at least one of the following: obtaining the road shape information of the vehicle object; obtaining a point trace processing identifier that matches the historical point trace set collected in the previous frame, wherein the point trace processing identifier is used to indicate the processing method of the first point trace set in the current frame; obtaining the number of historical point traces of the historical point trace object included in the historical point trace set.

[0197] As an optional scheme, the above-mentioned recognition device based on point trace set also includes: a condition detection unit, which is used to determine that the above-mentioned driving state information of the above-mentioned vehicle object does not meet the above-mentioned target measurement point condition when the above-mentioned road shape information indicates that the above-mentioned vehicle is in a non-straight driving state; when the above-mentioned road shape information indicates that the above-mentioned vehicle is in a straight driving state and the above-mentioned point trace processing identifier is a first value, determine that the above-mentioned driving state information of the above-mentioned vehicle object meets the above-mentioned target measurement point condition; when the above-mentioned road shape information indicates that the above-mentioned vehicle is in a straight driving state, and the weighted sum of the above-mentioned first point trace number and the above-mentioned historical point trace number is greater than a reference number threshold, determine that the above-mentioned driving state information of the above-mentioned vehicle object meets the above-mentioned target measurement point condition.

[0198] As an optional scheme, the above-mentioned second determination module also includes at least one of the following: when the ratio between the number of reference points of the reference point trace object and the second number of second points of the above-mentioned second point trace object is greater than a first threshold, the point trace processing identifier of the above-mentioned current frame is determined as the first value, wherein the above-mentioned reference point trace object is a point trace object outside the boundary of the area of ​​interest; when the ratio between the above-mentioned reference point trace number of the above-mentioned reference point trace object and the above-mentioned second number of points of the above-mentioned second point trace object is greater than a second threshold, the above-mentioned point trace processing identifier of the above-mentioned current frame is determined as the above-mentioned first value.

[0199] As an optional scheme, the above-mentioned second determination module is also used to: modify the value of the inter-frame counter matching the current frame to a first count value, wherein the above-mentioned first count value is greater than 0, and the value of the above-mentioned inter-frame counter is used to indicate the number of frames for performing the above-mentioned first filtering operation; when the above-mentioned driving status information of the above-mentioned self-vehicle object does not meet the above-mentioned target measurement point conditions, the first count value matching the above-mentioned inter-frame counter is reduced by 1.

[0200] As an optional scheme, the above-mentioned first acquisition unit 902 is used to: obtain the first point track set collected by the above-mentioned self-vehicle object in the current frame; when the driving state information of the above-mentioned self-vehicle object meets the above-mentioned target measurement point condition, and the number of the above-mentioned first point tracks of the above-mentioned first point track object is greater than the above-mentioned first quantity threshold, perform a first screening operation on the above-mentioned first point track set to obtain the above-mentioned candidate point track set, wherein the number of candidate point tracks of the candidate point track object included in the above-mentioned candidate point track set is less than or equal to the second quantity threshold, and the above-mentioned second quantity threshold is greater than the above-mentioned first quantity threshold.

[0201] As an optional scheme, the above-mentioned first acquisition unit 902 is also used for: when the driving status information of the above-mentioned vehicle object does not meet the above-mentioned target measurement point conditions, and the above-mentioned first point trace quantity of the above-mentioned first point trace object is greater than the above-mentioned first quantity threshold, performing a second screening operation on the above-mentioned first point trace set to obtain a third point trace set, wherein the number of third points of the third point trace object included in the above-mentioned third point trace set is less than or equal to the above-mentioned first quantity threshold; and performing an object recognition operation according to the above-mentioned third point trace set.

[0202] For a specific embodiment, reference may be made to the example shown in the above-mentioned recognition method based on a point trace set, which will not be described in detail in this example.

[0203] like Fig.10 In the electronic device shown, the memory 1002 can be used to store software programs and modules, such as the program instructions / modules corresponding to the point trace set-based recognition method and device in the embodiment of the present invention. The processor 1004 executes various functional applications and data processing by running the software programs and modules stored in the memory 1002, that is, to implement the above-mentioned point trace set-based recognition method. The memory 1002 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 1002 may further include a memory remotely located relative to the processor 1004, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. Among them, the memory 1002 can be specifically, but not limited to, used to store file information such as target files. As an example, such as Fig.10 As shown, the memory 1002 may include, but is not limited to, the first acquisition unit 902, the second acquisition unit 904, the determination unit 906, and the recognition unit 908 in the recognition device based on a point trace set. In addition, other module units in the recognition device based on a point trace set may also be included but are not limited to, which will not be repeated in this example.

[0204] Optionally, the transmission device 1006 is used to receive or send data via a network. Specific examples of the above-mentioned network may include a wired network and a wireless network. In one example, the transmission device 1006 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices and routers via a network cable so as to communicate with the Internet or a local area network. In one example, the transmission device 1006 is a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0205] In addition, the electronic device mentioned above further includes: a display 1008, and a connection bus 1010, which is used to connect various module components in the electronic device mentioned above.

[0206] According to one aspect of the present application, a computer program product is provided, the computer program product comprising a computer program / instruction, the computer program / instruction comprising a program code for executing the method shown in the flow chart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part, and / or installed from a removable medium. When the computer program is executed by a central processing unit, various functions provided by the embodiments of the present application are executed.

[0207] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0208] It should be noted that the computer system of the electronic device is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0209] In particular, according to an embodiment of the present application, the process described in each method flow chart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer readable medium, and the computer program contains a program code for executing the method shown in the flow chart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part, and / or installed from a removable medium. When the computer program is executed by a central processing unit, various functions defined in the system of the present application are executed.

[0210] According to one aspect of the present application, a computer-readable storage medium is provided, and a processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the methods provided in the above-mentioned various optional implementations.

[0211] Optionally, in this embodiment, the computer-readable storage medium may be configured to store a computer program for performing the following steps:

[0212] S1, obtaining driving state information of the vehicle object and a candidate point track set corresponding to the current frame, wherein the candidate point track set is obtained by screening a first point track set collected by the vehicle in the current frame, and the first point track set includes a plurality of first point track objects that pass the constant false alarm detection;

[0213] S2, when the driving state information of the self-vehicle object meets the target measurement point condition and the number of first point traces of the first point trace object is greater than a first number threshold, obtaining region boundary description information of the region of interest matching the current frame;

[0214] S3, according to the position information corresponding to each of the plurality of candidate point trace objects, determining the plurality of candidate point trace objects located in the region of interest as a plurality of second point trace objects, wherein the number of second point traces of the plurality of second point trace objects is less than or equal to the first number threshold;

[0215] S4, performing an object recognition operation according to a second point trace set including a plurality of the second point trace objects.

[0216] Optionally, in this embodiment, a person of ordinary skill in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by instructing hardware related to the electronic device through a program, and the program may be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0217] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0218] If the integrated units in the above embodiments are implemented in the form of software functional units and sold or used as independent products, they can be stored in the above computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling one or more computer devices (which can be personal computers, servers or network devices, etc.) to execute all or part of the steps of the methods of each embodiment of the present application.

[0219] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0220] In the several embodiments provided in the present application, it should be understood that the disclosed user equipment can be implemented in other ways. Among them, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0221] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0222] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0223] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A recognition method based on a point trace set, characterized in that: include: Acquire driving state information of the vehicle object and a candidate point track set corresponding to the current frame, wherein the candidate point track set is obtained by a first screening operation based on a first point track set collected by the vehicle in the current frame, and the first point track set includes a plurality of first point track objects detected by constant false alarm; When the driving state information of the vehicle object meets the target measurement point condition and the number of first point traces of the first point trace object is greater than a first number threshold, obtaining region boundary description information of the region of interest matching the current frame; According to the position information corresponding to each of the plurality of candidate point trace objects in the candidate point trace set, the plurality of candidate point trace objects located in the region of interest are determined as a plurality of second point trace objects, wherein the number of second point traces of the plurality of second point trace objects is less than or equal to the first number threshold; An object recognition operation is performed according to a second point trace set including a plurality of the second point trace objects.

2. The method according to claim 1, characterized in that The step of determining the plurality of candidate point trace objects located in the region of interest as a plurality of second point trace objects comprises at least one of the following: In the case where the region boundary description information includes the plane coordinates of at least one region boundary point, a plurality of second point trace objects are determined from the plurality of candidate point trace objects according to a comparison result of the plane coordinates of the candidate point trace object with the plane coordinates of at least one region boundary point; In a case where the area boundary description information includes an array of boundary point descriptions corresponding to at least one area boundary point, a plurality of second point trace objects are determined from a plurality of candidate point trace objects based on a comparison result between the azimuth index identifier of the candidate point trace object and the reference distance unit index identifier included in the array of boundary point descriptions, wherein the array of boundary point descriptions is used to indicate the correspondence between the boundary distance of the boundary point and the azimuth index.

3. The method according to claim 2, characterized in that The acquiring of the region boundary description information of the region of interest matching the current frame comprises at least one of the following: In a case where the driving environment of the vehicle object includes a target enclosure object, determining the region boundary description information according to a position corresponding to the target enclosure object; In a case where the driving environment of the vehicle object does not include a target enclosure object, the region boundary description information of the region of interest matching the current frame is determined according to the historical boundary description information corresponding to at least one historical frame.

4. The method according to claim 2, characterized in that: According to the comparison result between the azimuth index identifier of the candidate point trace object and the reference distance unit index identifier included in the boundary point description array, a plurality of second point trace objects are determined from the plurality of candidate point trace objects, including: Traversing a plurality of the candidate point trace objects, and determining a reference distance unit index identifier matching the azimuth index identifier from the boundary point description array according to the azimuth index identifier of the currently traversed candidate point trace object; Determine a reference azimuth index identifier corresponding to the reference distance unit index identifier according to the boundary point description array; When the azimuth index identifier of the candidate point trace object matches the reference azimuth index identifier, the candidate point trace object currently traversed is determined as the second point trace object.

5. The method according to claim 2, characterized in that: The obtaining of the driving state information of the vehicle object includes at least one of the following: Obtaining road shape information on which the vehicle object is located; Acquire a point trace processing identifier that matches the historical point trace set collected in the previous frame, wherein the point trace processing identifier is used to indicate a processing method for the first point trace set in the current frame; Gets the number of historical points in the historical point collection.

6. The method according to claim 5, characterized in that After obtaining the driving state information of the vehicle object, the method further includes: When the road shape information indicates that the vehicle is in a non-straight driving state, determining that the driving state information of the vehicle object does not meet the target measurement point condition; When the road shape information indicates that the vehicle is in a straight-line driving state and the point trace processing identifier is a first value, determining that the driving state information of the vehicle object meets the target measurement point condition; When the road shape information indicates that the vehicle is in a straight-line driving state, and a weighted sum of the first point trace quantity and the historical point trace quantity is greater than a reference quantity threshold, it is determined that the driving state information of the vehicle object meets the target measurement point condition.

7. The method according to claim 5, characterized in that After determining the plurality of candidate point trace objects located in the region of interest as a plurality of second point trace objects according to the position information corresponding to each of the plurality of candidate point trace objects in the candidate point trace set, the method further includes at least one of the following: When the ratio between the first reference trace quantity of the reference trace object and the second trace quantity of the second trace object is greater than a first threshold, determining the trace processing identifier of the current frame as a first value, wherein the reference trace object is a trace object outside the boundary of the region of interest; When the ratio between the second reference point quantity of the reference point object and the second point quantity of the second point object is greater than a second threshold, the point processing identifier of the current frame is determined to be the first value.

8. The method according to claim 7, characterized in that After determining the trace processing identifier of the current frame as the first value, the method further includes: Modify the value of the inter-frame counter matching the current frame to a first count value, wherein the first count value is greater than 0, and the value of the inter-frame counter is used to indicate the number of frames for performing the first filtering operation; When the driving state information of the vehicle object does not satisfy the target measurement point condition, the first count value matched with the inter-frame counter is reduced by 1.

9. The method according to claim 1, characterized in that: The step of obtaining the driving state information of the vehicle object and the candidate point trace set corresponding to the current frame includes: Obtaining a first point trace set collected by the vehicle object in the current frame; When the driving state information of the vehicle object meets the target measuring point condition and the number of the first points of the first point trace object is greater than the first quantity threshold, the first screening operation is performed on the first point trace set to obtain the candidate point trace set, wherein the number of candidate points of the candidate point trace objects included in the candidate point trace set is less than or equal to a second quantity threshold, and the second quantity threshold is greater than the first quantity threshold.

10. The method according to claim 9, characterized in that After obtaining the driving state information of the vehicle object, the method further includes: When the driving state information of the self-vehicle object does not meet the target measurement point condition, and the number of the first point traces of the first point trace object is greater than the first number threshold, performing a second screening operation on the first point trace set to obtain a third point trace set, wherein the number of third point traces of the third point trace object included in the third point trace set is less than or equal to the first number threshold; An object recognition operation is performed according to the third point trace set.

11. A recognition device based on a point trace set, characterized in that: include: A first acquisition unit is used to acquire driving state information of the ego vehicle object and a candidate point track set corresponding to the current frame, wherein the candidate point track set is obtained by a first screening operation based on a first point track set collected by the ego vehicle in the current frame, and the first point track set includes a plurality of first point track objects that pass the constant false alarm detection; A second acquisition unit is used to acquire the region boundary description information of the region of interest matching the current frame when the driving state information of the self-vehicle object meets the target measurement point condition and the number of first point traces of the first point trace object is greater than a first number threshold; a determining unit, configured to determine the plurality of candidate point trace objects located in the region of interest as a plurality of second point trace objects according to the position information corresponding to each of the plurality of candidate point trace objects in the candidate point trace set, wherein the number of second point traces of the plurality of second point trace objects is less than or equal to the first number threshold; The recognition unit is used to perform an object recognition operation according to a second point trace set including a plurality of the second point trace objects.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program executes the method described in any one of claims 1 to 10 when executed by an electronic device.

13. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 10 through the computer program.

14. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

Images