Target classification method and device

Abstract

The invention provides a target classification method and device and electronic equipment, and the method comprises the steps: obtaining a motion track point set of a target object, determining a motion interval angle corresponding to the target object based on the motion track point set, and determining a first motion state feature of the target object based on the motion interval angle and the motion track point set, and determining the category of the target object according to the first motion state feature. By adopting the method, the accuracy of target classification is improved.

Description

Technical Field

[0001] The present disclosure relates to the field of image processing technology, and in particular to a target classification method and device. Background Art

[0002] How to classify pedestrians and vehicles is an important issue in the application of vehicle-mounted millimeter-wave radar. Traditional target classification methods rely on information processing methods such as cameras and lidar. These classification methods have complex algorithm processes, high requirements on radar operating performance, and long overall calculation time, resulting in low efficiency and accuracy of target classification. Therefore, how to improve the accuracy of target classification has become a technical problem that needs to be solved urgently. Summary of the invention

[0003] The present disclosure provides a target classification method, device, electronic device and storage medium to at least solve the above technical problems existing in the prior art.

[0004] According to a first aspect of the present disclosure, a target classification method is provided, the method comprising:

[0005] Get the target object's motion track point set;

[0006] Based on the set of motion track points, determining a motion interval angle corresponding to the target object;

[0007] Determining a first motion state feature of the target object based on the motion interval angle and the motion track point set;

[0008] The category of the target object is determined according to the first motion state feature.

[0009] In one possible implementation, determining the motion interval angle corresponding to the target object based on the motion track point set includes:

[0010] The following formula is used to determine the motion interval angle corresponding to the target object based on the motion track point set:

[0011]

[0012] in, is the motion interval angle corresponding to the target object, R is the motion distance of the target object, and r is the maximum distance between the track points corresponding to the target object.

[0013] In one possible implementation, determining the first motion state feature of the target object based on the motion interval angle and the motion track point set includes:

[0014] Determine the track point interval corresponding to the movement interval angle;

[0015] Calculating the Doppler velocity characteristics between each track point in the track point interval;

[0016] Determining whether there is a Doppler velocity feature satisfying a preset Doppler condition among the plurality of Doppler velocity features;

[0017] If so, determining the feature of the target object satisfying the preset Doppler condition as the first motion state feature of the target object;

[0018] If not, the feature that the target object does not satisfy the preset Doppler condition is determined as the first motion state feature of the target object.

[0019] In one possible implementation manner, the calculating the Doppler velocity characteristics between each track point in the track point interval includes:

[0020] The Doppler velocity characteristics between each track point in the track point interval are calculated using the following formula:

[0021]

[0022] Among them, th is the Doppler velocity characteristic between two track points in the track point interval, = , is the Doppler velocity of track point 1 within the track point interval, is the Doppler velocity of track point 2 within the track point interval, , , is the lateral movement speed of the target object, is the longitudinal motion speed of the target object, is the motion range angle, It is the difference between the azimuths of two track points in the track point interval.

[0023] In one possible implementation manner, determining the category of the target object according to the first motion state feature includes:

[0024] Determine an initial classification result corresponding to the target object based on the motion track point set;

[0025] A target category of the target object is determined based on the initial classification result and the first motion state feature.

[0026] In one possible implementation manner, determining the initial classification result corresponding to the target object based on the set of moving track points includes:

[0027] Acquire a first preset number of continuous motion track frames corresponding to the target object;

[0028] Determine the target object size information and the number of motion track points corresponding to each motion track frame;

[0029] Determine whether the number of motion track frames in which the target object size information meets a preset size threshold and the number of motion track is greater than a preset number of track is greater than a second preset number;

[0030] If not, determining the movement track confidence and movement distance corresponding to the target object;

[0031] Based on the motion track confidence and the motion distance, an initial classification result corresponding to the target object is determined.

[0032] In one embodiment, the method further comprises:

[0033] If the target object size information meets the preset size threshold and the number of motion track frames with a motion point number greater than the preset point track number is greater than the second preset number, the category of the target object is determined to be the first category, and the first category is used as the initial classification result.

[0034] In one embodiment, the method further comprises:

[0035] The determining, based on the motion track confidence and the motion distance, an initial classification result corresponding to the target object includes:

[0036] If the movement distance satisfies a preset distance threshold and the movement track confidence is less than a first preset confidence threshold, determining that the category of the target object is a second category, and taking the second category as an initial classification result;

[0037] If the motion track confidence is greater than a second preset confidence threshold, determining that the category of the target object is the third category, and using the third category as an initial classification result;

[0038] If the motion track confidence is greater than or equal to the first preset confidence threshold and less than or equal to the second preset confidence threshold, the category of the target object is determined to be the fourth category, and the fourth category is used as the initial classification result. The first category, the second category, the third category and the fourth category are all different.

[0039] In one possible implementation, determining the target category of the target object based on the initial classification result and the first motion state feature includes:

[0040] Based on the initial classification result, obtaining the moving speed of the target object;

[0041] The target category of the target object is determined according to the first motion state feature, the motion speed and the motion track confidence.

[0042] According to a second aspect of the present disclosure, a target classification device is provided, the device comprising:

[0043] A track information acquisition module is used to obtain a set of moving track points of a target object;

[0044] An interval angle determination module is used to determine the movement interval angle corresponding to the target object based on the movement track point set;

[0045] A feature determination module, configured to determine a first motion state feature of the target object based on the motion interval angle and the motion track point set;

[0046] A classification module is used to determine the category of the target object according to the first motion state feature.

[0047] According to a third aspect of the present disclosure, there is provided an electronic device, including:

[0048] at least one processor; and

[0049] a memory communicatively connected to the at least one processor; wherein,

[0050] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method described in the present disclosure.

[0051] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause the computer to execute the method described in the present disclosure.

[0052] The target classification method and device disclosed in the present invention obtain a set of moving track points of a target object, determine the moving interval angle corresponding to the target object based on the moving track point set, determine the first moving state feature of the target object based on the moving interval angle and the moving track point set, and determine the category of the target object according to the first moving state feature. That is, the first moving state feature of the target object can be extracted through the moving track point set of the target object without relying on a radar with too high performance, and the category of the target object can be analyzed using the first moving state feature, thereby improving the accuracy of target classification.

[0053] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The above and other objects, features and advantages of the exemplary embodiments of the present disclosure will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present disclosure are shown in an exemplary and non-limiting manner, in which:

[0055] In the drawings, the same or corresponding reference numerals represent the same or corresponding parts.

[0056] Figure 1 A schematic diagram of a process flow of a target classification method provided by an embodiment of the present disclosure is shown;

[0057] Figure 2 A schematic diagram of a motion interval angle of a target object provided by an embodiment of the present disclosure is shown;

[0058] Figure 3 A schematic diagram of a process for determining a first motion state feature of a target object provided by an embodiment of the present disclosure is shown;

[0059] Figure 4 A schematic diagram of Doppler velocity characteristics of a target object provided by an embodiment of the present disclosure is shown;

[0060] Figure 5 A schematic diagram showing Doppler velocity characteristics of another target object provided by an embodiment of the present disclosure is shown;

[0061] Figure 6 A schematic diagram of an object category determination process provided by an embodiment of the present disclosure is shown;

[0062] Figure 7 A schematic diagram of the structure of a target classification device provided by an embodiment of the present disclosure is shown;

[0063] Figure 8 A schematic diagram of the structure of an electronic device provided by an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0064] In order to make the purpose, features, and advantages of the present disclosure more obvious and easy to understand, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present disclosure.

[0065] Since the traditional target classification method does not effectively utilize the target's track information and the algorithm process is complex, the radar operating performance requirements are high, and the overall calculation is time-consuming, resulting in low efficiency and low accuracy of target classification. Therefore, in order to improve the accuracy of target classification, the present disclosure provides a target classification method, device, electronic device and storage medium. The target classification method provided by the present disclosure can be applied to any electronic device that can perform image processing, including but not limited to computers, mobile phones and tablets. The target classification method provided by the present disclosure can be applied to a vehicle-mounted system to classify targets encountered during vehicle driving.

[0066] The technical solution of the embodiment of the present disclosure will be described below in conjunction with the drawings in the embodiment of the present disclosure.

[0067] Figure 1 A schematic diagram of a process flow of a target classification method provided by an embodiment of the present disclosure is shown. Figure 1 As shown, the method includes:

[0068] S101, obtaining a set of movement track points of a target object.

[0069] In the present disclosure, the target objects may be pedestrians, trucks, cars, VRUs (Vulnerable Road Users), animals, etc. encountered by the vehicle during driving, wherein VRUs may include pedestrians, cyclists, motorcyclists, and people taking public transportation, etc.

[0070] In the present disclosure, a radar on a vehicle can be used to collect radar information of objects encountered by the vehicle during driving, and a motion track of the object can be obtained based on the radar information, and a set of the motion track of the object can be used as a set of motion track points of the object.

[0071] The movement track point of the target object may include the location information, movement speed information, movement direction information, etc. of the target object.

[0072] S102: Determine a motion interval angle corresponding to the target object based on the motion track point set.

[0073] In the present disclosure, the motion interval angle corresponding to the target object is the angle that represents the running direction of the target object. For example, Figure 2 A schematic diagram of a motion interval angle of a target object provided by an embodiment of the present disclosure is shown. Figure 2 As shown, vehicle 201 encounters target object A during driving, and target object A generates track point 202 during its movement. Target object A takes vehicle 201 as a reference, and its running direction range is within the range of interval angle α, and the track point of target object A is also within the range of interval angle α.

[0074] In a possible implementation, the determining the movement interval angle corresponding to the target object based on the movement track point set may include: using the following formula to determine the movement interval angle corresponding to the target object based on the movement track point set:

[0075]

[0076] in, is the motion interval angle corresponding to the target object, R is the motion distance of the target object, and r is the maximum distance between the track points corresponding to the target object.

[0077] by Figure 2 Take the track point 202 as an example. Figure 2 As shown, during the movement of the target object A, a track point 202 is generated, and the movement distance of the track point 202 is L1. Among all the track points generated during the movement of the target object A, the track point with the largest distance from the track point 202 is the track point max, and the distance between the track point max and the track point 202 is Lmax. Then, the movement interval angle corresponding to the target object can be calculated. for .

[0078] In the present disclosure, the motion interval angle corresponding to the target object is greater than the first preset angle threshold and less than the second preset angle threshold, wherein the first preset angle threshold can generally be set to 1° or 1.5°, and the second preset angle threshold can generally be set to 5° or 5.5°.

[0079] In another possible implementation, if according to the formula “ The calculated motion interval angle is less than or equal to the first preset angle threshold or greater than or equal to the second preset angle threshold, and any angle value between the first preset angle threshold and greater than or equal to the second preset angle threshold can be determined as the motion interval angle corresponding to the target object.

[0080] S103: Determine a first motion state feature of the target object based on the motion interval angle and the motion track point set.

[0081] In the present disclosure, the Doppler feature of the target object may be determined as the first motion state feature.

[0082] S104: Determine the category of the target object according to the first motion state feature.

[0083] In the present disclosure, the categories of target objects may be vulnerable road users (VRUs) such as pedestrians, bicycles, and motorcycles.

[0084] By adopting the target classification method disclosed in the present invention, a set of moving track points of a target object is obtained, and based on the set of moving track points, the moving interval angle corresponding to the target object is determined, and based on the moving interval angle and the set of moving track points, the first moving state feature of the target object is determined, and the category of the target object is determined according to the first moving state feature. That is, without relying on a radar with too high performance, the first moving state feature of the target object can be extracted through the moving track point set of the target object, and the category of the target object can be analyzed using the first moving state feature, thereby improving the accuracy of target classification.

[0085] In one possible implementation, Figure 3 FIG. 2 shows a schematic diagram of a process for determining a first motion state feature of a target object provided by an embodiment of the present disclosure. Figure 3 As shown, the determining of the first motion state feature of the target object based on the motion interval angle and the motion track point set includes:

[0086] S301, determining a track point interval corresponding to the movement interval angle.

[0087] In the present disclosure, for a target object, a reference vehicle can be used as the interval origin, and the area range with an angle corresponding to the movement interval of the target object can be determined as the track point interval of the target object, and the track point interval includes the current location of the target object. The reference vehicle can be the vehicle that the user is driving. The target object is an object encountered by the reference vehicle during driving.

[0088] S302, calculating the Doppler velocity characteristics between each track point in the track point interval.

[0089] In the present disclosure, for each target object, the Doppler velocity feature between any two track points in the track point interval corresponding to the target object can be calculated to obtain multiple Doppler velocity features corresponding to the target object. Alternatively, for each target object, the Doppler velocity feature between every two adjacent track points in the track point interval corresponding to the target object can be calculated to obtain multiple Doppler velocity features corresponding to the target object.

[0090] In a possible implementation, the following formula is used to calculate the Doppler velocity characteristics between each track point in the track point interval:

[0091]

[0092] Among them, th is the Doppler velocity characteristic between two track points in the track point interval, = , is the Doppler velocity of track point 1 within the track point interval, is the Doppler velocity of track point 2 within the track point interval, , , is the lateral movement speed of the target object, is the longitudinal motion speed of the target object, is the motion range angle, It is the difference between the azimuths of two track points in the track point interval.

[0093] In the present disclosure, if the number of track points associated with the target object is not greater than 1, the target object has no corresponding Doppler velocity feature. If the target object is not in motion, the target object has no corresponding Doppler velocity feature.

[0094] S303: Determine whether there is a Doppler velocity feature that meets a preset Doppler condition among the multiple Doppler velocity features.

[0095] In the present disclosure, the preset Doppler condition may be a condition for screening the Doppler characteristics of pedestrians. The Doppler characteristics of pedestrians characterize the motion information related to the Doppler effect during the movement of pedestrians. The preset Doppler condition may include: a Doppler velocity feature having a corresponding characteristic value greater than a preset characteristic value exists in a plurality of the Doppler velocity characteristics. If at least one Doppler velocity feature having a characteristic value greater than a preset characteristic value exists in a plurality of Doppler velocity characteristics corresponding to the target object, it may be determined whether there is a Doppler velocity feature in the Doppler velocity characteristics that satisfies the preset Doppler condition. The preset characteristic value may be set as required, for example, the preset characteristic value may be set to 2 or 3, etc.

[0096] S304: If yes, determine the feature of the target object satisfying the preset Doppler condition as the first motion state feature of the target object.

[0097] In the present disclosure, the parameter Microfalg can be set to characterize whether the target object meets the preset Doppler condition. If the target object meets the characteristics of the preset Doppler condition, the value of Microfalg can be set to 1. If the target object does not meet the characteristics of the preset Doppler condition, the value of Microfalg can be set to 0.

[0098] If at least one Doppler velocity feature among the multiple Doppler velocity features corresponding to the target object has a feature value greater than a preset feature value, the feature of the target object that meets the preset Doppler condition can be used as the first motion state feature of the target object. For example, the feature Microfalg=1 can be used as the first motion state feature of the target object.

[0099] In the present disclosure, a feature of a target object that satisfies a preset Doppler condition is determined as a first motion state feature of the target object. If the preset Doppler condition is a condition set based on the Doppler speed of a pedestrian, then the first motion state feature is a feature that characterizes that the target object may be a pedestrian. If the preset Doppler condition is a condition set based on the Doppler speed of a bicycle, then the first motion state feature is a feature that characterizes that the target object may be a bicycle.

[0100] S305: If not, determining the feature of the target object not satisfying the preset Doppler condition as the first motion state feature of the target object.

[0101] If there is no Doppler velocity feature whose characteristic value is greater than the preset characteristic value among the multiple Doppler velocity features corresponding to the target object, the feature that the target object does not meet the preset Doppler condition can be used as the first motion state feature of the target object. For example, the feature Microfalg=0 can be used as the first motion state feature of the target object.

[0102] In the present disclosure, the Doppler velocity feature of the target object represents the frequency change caused by the movement of the target object, and the Doppler velocity feature of the target object can reflect the speed of the target object relative to the reference vehicle. Figure 4 FIG. 4 shows a schematic diagram of a Doppler velocity characteristic of a target object provided by an embodiment of the present disclosure, such as Figure 4 As shown, the target object 401 is a pedestrian. Figure 4 The horizontal axis of the middle coordinate represents time in seconds. Figure 4 The vertical axis represents the Doppler velocity of the target object 401 , in meters per second. Figure 5 FIG. 2 shows another schematic diagram of Doppler velocity characteristics of a target object provided by an embodiment of the present disclosure. Figure 5 As shown, the target object 501 is a bicycle. Figure 5 The horizontal axis of the middle coordinate represents time in seconds. Figure 5 The vertical axis represents the Doppler velocity of the target object 501, in meters per second. Figure 4 The Doppler velocity distribution of pedestrians shown in Figure 5 From the Doppler velocity distribution of bicycles shown in , it can be seen that pedestrians have a longer distribution in the Doppler velocity direction compared to bicycles.

[0103] Due to the different Doppler velocity distributions of different objects, in this step, the Doppler velocity features of the target are extracted. By presetting the Doppler conditions and determining the first motion state features that can reflect that the target may be a pedestrian or a bicycle, the accuracy of classifying the target object using the first motion state features of the target object can be improved.

[0104] In one possible implementation, Figure 6 FIG. 2 shows a schematic diagram of an object category determination process provided by an embodiment of the present disclosure. Figure 6 As shown, determining the category of the target object according to the first motion state feature includes:

[0105] S601: Determine an initial classification result corresponding to the target object based on the motion track point set.

[0106] In one possible implementation, the determining of the initial classification result corresponding to the target object based on the set of moving track points may include steps A1-A5:

[0107] Step A1, obtaining a first preset number of continuous motion track frames corresponding to the target object.

[0108] The first preset number can be set to 8 or 10, etc. In the present disclosure, the radar sensor on the reference vehicle can collect radar information of the target object, and the radar information includes a motion track frame of the target object. The motion track frame includes attribute information of the target object and attribute information of the motion point track of the target object. The attribute information of the motion point track of the target object includes the position coordinates, lateral speed, longitudinal speed, etc. of the motion point track.

[0109] In the present disclosure, whether the target object is a dynamic target can be determined based on the lateral speed and longitudinal speed of the target object. Specifically, if the longitudinal speed and lateral speed of the track point of the target object in the current track frame are both greater than the target speed threshold, the target object can be determined to be a dynamic target, and the target object can be further classified. If the longitudinal speed or lateral speed of the track point of the target object in the current track frame is not greater than the target speed threshold, the target object can be determined to be stationary, and the classification judgment of the target object can use the track information before the target object turns to a stationary state.

[0110] In the present disclosure, for an obscured target object, some track points of the target object may not be obtained. In this regard, a plurality of track frames may be set in the present disclosure, and the number of track frames may be set to 8 or 10, etc., according to actual needs. The track point information of the target object when it passes through the track frame is collected using the track frame. The ID of the track frame is associated with the track point information of the target object in the track frame and stored. For each target object, if the track of the target object calculated using the motion track frame does not match the track of the target object stored in the track frame associated with the target object, the motion track frame of the target object may be marked with an occlusion mark, indicating that the target object is obscured at the position of the motion track frame.

[0111] Step A2, determining the target object size information and the number of motion track points corresponding to each motion track frame.

[0112] In the present disclosure, for each motion track frame of each target object, the length, width and number of motion track points of the target object corresponding to the motion track frame can be calculated. The motion track points are the track points of the target object.

[0113] In the present disclosure, the SVM trainer and motion track frame obtained through Matlab offline training can be used to determine the target object size information and the number of motion point tracks corresponding to each motion track frame. Specifically, the information of the motion track frame can be input into the SVM trainer to obtain the track RCS (Radar Cross Section), track frame length and track frame width and other information corresponding to the motion track frame output by the SVM trainer. The SVM trainer can also output the weights of the track RCS (Radar Cross Section), track frame length and track frame width and other information corresponding to the motion track frame. In the present disclosure, if the current track of the target object is blocked, the current track RCS value can be transferred, and the transferred value can be used to classify and judge the target object. If the current track of the target object crosses the track, the current track RCS value can be reduced.

[0114] In the present disclosure, the length and width of the target object can be determined according to the track RCS, track frame length and track frame width corresponding to the motion track frame, and the weight of each information. In a possible implementation, the track frame length and track frame width can also be used as the length and width of the target object, respectively. According to the length and width of the target object.

[0115] Step A3, determining whether the target object size information meets a preset size threshold and the number of motion track frames in which the number of motion track is greater than a preset number of track is greater than a second preset number.

[0116] The second preset number can be set according to application requirements, for example, set to 3 or 4, etc. The preset size threshold may include a preset width threshold and / or a preset length threshold. The preset width threshold and / or the preset length threshold can be set according to application requirements, for example, the preset width threshold can be set to 3 meters or 4 meters, etc., and the preset length threshold can be set to 10 meters or 15 meters, etc. The preset number of point traces can be set to 2 or 3, etc.

[0117] Step A4: if not, determine the movement track confidence and movement distance corresponding to the target object.

[0118] In the present disclosure, taking a large truck as an example, the size of a large truck is larger than that of a pedestrian or a two-wheeled vehicle, and a size threshold can be set to identify whether the target is a large truck using the size information of the target. For example, the preset size threshold can be set as a threshold representing the size of a large truck, and the preset size threshold can be used to determine whether the target is a large truck.

[0119] If the target object size information does not meet the preset size threshold, or the number of motion track frames with a motion point number greater than the preset point track number is not greater than a second preset number, that is, the target object cannot be a large truck, but a smaller target. At this time, it is difficult to continue to identify the category of the target object through the size information, and it is necessary to further determine the motion track confidence and motion distance corresponding to the target object, and use the motion track confidence and motion distance corresponding to the target object to determine the initial classification result corresponding to the target object.

[0120] In the present disclosure, a Bayesian multi-frame cumulative confidence algorithm can be used to determine the confidence of the motion track corresponding to the target object. Specifically, the prior probability of the motion track frame can be determined by the confidence of the motion track frame of the historical data or the previous frame, and then the likelihood of the target detection result of the motion track frame is calculated. The posterior probability of the motion track frame is determined by using the Bayesian theorem, the prior probability and the likelihood of the target detection result of the motion track frame, and the posterior probabilities after multiple frame updates are superimposed to obtain the confidence of the current motion track frame. The confidence of the motion track frame is used to characterize the credibility of the target classification result.

[0121] In the present disclosure, if the current motion state of the target object is a motion point track that is not associated with the target object, the motion track confidence of the current motion track frame can be calculated based on the motion track confidence of the motion track frames of the previous frames. For example, the average of the motion track confidences of the motion track frames of the previous frames can be calculated as the motion track confidence of the current motion track frame, or the motion track confidence of the previous frame of the motion track frame can be maintained as the motion track confidence of the current motion track frame.

[0122] In the present disclosure, if the current motion state of the target object is associated with the motion track of the target object, if the type of the target object represented by the current motion track frame is the same as the type of the target object represented by the previous motion track frame, the motion track confidences of the previous motion track frames can be accumulated, and the obtained sum value is used as the motion track confidence of the current motion track frame;

[0123] If the type of the target object represented by the current motion track frame is different from the type of the target object represented by the previous motion track frame, the motion track confidence of the motion track frames of the previous frames can be reduced, and the largest motion track confidence among the motion track confidences of the previous frames of motion track frames obtained after the reduction processing is used as the motion track confidence of the current motion track frame.

[0124] In the present disclosure, the movement distance corresponding to the target object can be calculated by the radar of the reference vehicle. The movement distance corresponding to the target object can include a lateral movement distance and a longitudinal movement distance, and the lateral movement distance and the longitudinal movement distance are determined by taking the coordinate system corresponding to the reference vehicle as the coordinate system.

[0125] In the present disclosure, the motion track confidence of the target object can be evaluated by using the motion track measurement value of the target object and the radar sensor error characteristics of the reference vehicle. Alternatively, a filtering algorithm can be used to evaluate the motion track confidence of the target object based on the motion track information of the target object.

[0126] Step A5: determining an initial classification result corresponding to the target object based on the motion track confidence and the motion distance.

[0127] In the present disclosure, the initial classification result may be a VRU, a car, or an unknown category.

[0128] In a possible implementation, to determine the initial classification result corresponding to the target object, step B1 may also be performed, if the size information of the target object meets the preset size threshold and the number of motion track frames is greater than the preset number of motion track frames and is greater than the second preset number, determine that the category of the target object is the first category, and use the first category as the initial classification result.

[0129] In the present disclosure, a size threshold can be set to identify whether the target is a large truck using the size information of the target. If the size information of the target object meets the preset size threshold and the number of motion track frames with a motion point number greater than the preset number of motion track frames is greater than a second preset number, that is, the size of the target object is large enough, and a plurality of consecutive motion track frames reflect that the size of the target object is large enough, then it can be determined that the target object is a large truck.

[0130] In the present disclosure, the first category is a large truck. If the second preset number is set to 3, when the target object size information meets the preset size threshold and the number of motion track frames in which the number of motion track is greater than the preset number of track is greater than 3, that is, in the continuous multiple frames of motion track frames, there are 3 frames of motion track frames in which the target object size information meets the preset size threshold and the number of motion track is greater than the preset number of track, then it can be determined that the target object is a large truck.

[0131] In the present disclosure, if it is determined that the initial classification result of the target object is the first category, the first category may be determined as the target category of the target object, that is, the first category may be determined as the final classification result of the target object.

[0132] In another possible implementation, the determining the initial classification result corresponding to the target object based on the motion track confidence and the motion distance may include steps C1-C4:

[0133] Step C1: if the movement distance meets a preset distance threshold and the movement track confidence is less than a first preset confidence threshold, determine that the category of the target object is a second category, and use the second category as an initial classification result.

[0134] In the present disclosure, the preset distance threshold includes a preset lateral distance threshold and / or a preset longitudinal distance threshold. The preset longitudinal distance threshold can be set according to the actual application scenario, for example, set to 3 meters or 4 meters, etc. The preset lateral distance threshold can be set according to the actual application scenario, for example, set to 20 meters or 21 meters, etc. In the present disclosure, if the lateral movement distance corresponding to the target object is less than the preset lateral distance threshold and the longitudinal movement distance corresponding to the target object is less than the preset longitudinal distance threshold, it is determined that the movement distance meets the preset distance threshold. The first preset confidence threshold can be set to 0.5 or 0.52, etc.

[0135] For example, if the preset longitudinal distance threshold is set to 3 meters and the preset lateral distance threshold is set to 20 meters, if the lateral movement distance corresponding to the target object is less than 3 meters and the longitudinal movement distance corresponding to the target object is less than 20 meters, in a single-frame motion track frame, the movement distance of the target object is greater than that of VRUs such as pedestrians and two-wheeled vehicles but does not reach the distance of trucks and cars. These features cannot be used to classify the target object into VRUs, trucks, and cars. Therefore, an unknown category can be set to count such situations.

[0136] The second category may be an unknown category. For example, if the preset lateral distance threshold is set to 20 meters, the preset longitudinal distance threshold is set to 3 meters, and the first preset confidence threshold is set to 0.5. If the lateral distance of the target object is less than 20 meters, the longitudinal distance of the target object is less than 3 meters, and the motion track confidence of the target object is less than 0.5, it can be determined that the category of the target object is an unknown category.

[0137] Step C2: if the motion track confidence is greater than a second preset confidence threshold, determine that the category of the target object is the third category, and use the third category as an initial classification result.

[0138] The second preset confidence threshold may be set to 0.6 or 0.65, etc. The second preset confidence threshold is greater than the first preset confidence threshold. The third category is cars.

[0139] In the present disclosure, if the movement distance of the target object meets a preset distance threshold and the movement track confidence is greater than a second preset confidence threshold, it can be determined that the target object is a car.

[0140] In the present disclosure, if it is determined that the initial classification result of the target object is the third category, the third category may be determined as the target category of the target object, that is, the third category may be determined as the final classification result of the target object.

[0141] Step C3, if the motion track confidence is greater than or equal to the first preset confidence threshold and less than or equal to the second preset confidence threshold, determine that the category of the target object is the fourth category, and use the fourth category as the initial classification result. The first category, the second category, the third category and the fourth category are all different.

[0142] In the present disclosure, the fourth category is VRU. If the moving distance of the target object meets the preset distance threshold and the moving track confidence is greater than or equal to the first preset confidence threshold and less than or equal to the second preset confidence threshold, the target object can be determined to be a VRU.

[0143] In the present disclosure, there are also large differences in the movement speeds of objects of different categories. Generally, the movement speed of a car is greater than that of a two-wheeled vehicle, and the movement speed of a two-wheeled vehicle is greater than that of a pedestrian. Therefore, when the target object size information does not meet the preset size threshold or the number of motion track frames with a motion point number greater than the preset point track number is not greater than a second preset number, the movement speed of the target object corresponding to each motion track frame can also be obtained, and the target object can be classified using the movement speed of the target object. For example, if the classification result of the target object corresponding to the previous motion track frame is a pedestrian category, and the movement speed of the target object corresponding to the current motion track frame is less than the preset speed threshold, the category of the target object corresponding to the current motion track frame can be determined as a pedestrian. The preset speed threshold can be set to a threshold that can reflect the movement of pedestrians according to application requirements, for example, set to 2.5m / s or 2.6m / s, etc.

[0144] S602: Determine a target category of the target object based on the initial classification result and the first motion state feature.

[0145] In the present disclosure, if the initial classification result is a VRU category, the movement speed of the target object and / or the classification result corresponding to the previous motion track frame can be obtained, and the category of the target object can be further determined using the movement speed of the target object and / or the classification result corresponding to the previous motion track frame and the first motion state characteristics of the target object.

[0146] In one embodiment, determining the target category of the target object based on the initial classification result and the first motion state feature may include steps D1-D2:

[0147] Step D1, based on the initial classification result, obtaining the movement speed of the target object.

[0148] In the present disclosure, the position and speed of the target object included in the radar data of the moving track frame can be analyzed, and the moving speed of the target object in each moving track frame can be calculated according to the position change and time interval of the target object.

[0149] Step D2, determining the target category of the target object according to the first motion state feature, the motion speed and the motion track confidence.

[0150] In the present disclosure, if the target object satisfies the following conditions a1 to a3, it can be determined that the target category of the target object is a pedestrian:

[0151] Condition a1: The target object's moving speed is less than a preset speed threshold;

[0152] Condition a2: The initial classification result of the target object is VRU;

[0153] Condition a3: The first motion state feature of the target object is a feature that the target object satisfies a preset Doppler condition.

[0154] In the present disclosure, if the initial classification result is VRU, the classification result corresponding to the previous motion track frame can also be obtained, so that the target category of the target object can be determined according to the motion speed of the target object, the classification result corresponding to the previous motion track frame and the motion track confidence. Specifically, in the present disclosure, if the target object meets the following conditions b1 to b4, the target category of the target object can be determined to be a two-wheeled vehicle, which can include motorcycles and bicycles, etc.:

[0155] Condition a1: The target object's moving speed is greater than a preset speed threshold;

[0156] Condition a2: The classification result corresponding to the previous motion track frame of the target object is a pedestrian;

[0157] Condition a3: the target object's motion track confidence is less than a first preset confidence threshold;

[0158] Condition a4: the first motion state characteristic of the target object is a characteristic that the target object does not satisfy a preset Doppler condition.

[0159] By adopting the target classification method disclosed in the present invention, the first motion state feature of the target object can be extracted through the motion track point set of the target object without relying on a radar with too high performance, and the category of the target object can be analyzed using the first motion state feature, thereby improving the accuracy of target classification.

[0160] Based on the same concept, according to the target classification method provided by the above embodiment of the present disclosure, correspondingly, another embodiment of the present disclosure also provides a target classification device, whose structural schematic diagram is shown in FIG. Figure 8As shown, specifically including:

[0161] Track information acquisition module 801, used to obtain a set of moving track points of a target object;

[0162] An interval angle determination module 802 is used to determine the movement interval angle corresponding to the target object based on the movement track point set;

[0163] A feature determination module 803, configured to determine a first motion state feature of the target object based on the motion interval angle and the motion track point set;

[0164] The classification module 804 is used to determine the category of the target object according to the first motion state feature.

[0165] By using the target classification device disclosed in the present invention, a set of moving track points of a target object is obtained, and based on the set of moving track points, the moving interval angle corresponding to the target object is determined, and based on the moving interval angle and the set of moving track points, the first moving state feature of the target object is determined, and the category of the target object is determined according to the first moving state feature. That is, without relying on a radar with too high performance, the first moving state feature of the target object can be extracted through the moving track point set of the target object, and the category of the target object can be analyzed using the first moving state feature, thereby improving the accuracy of target classification.

[0166] In one possible implementation, the interval angle determination module 802 is specifically configured to determine the motion interval angle corresponding to the target object based on the motion track point set using the following formula:

[0167]

[0168] in, is the motion interval angle corresponding to the target object, R is the motion distance of the target object, and r is the maximum distance between the track points corresponding to the target object.

[0169] In one possible implementation, the feature determination module 803 is specifically used to determine the track point interval corresponding to the motion interval angle; calculate the Doppler velocity characteristics between each track point in the track point interval; determine whether there is a Doppler velocity feature that meets a preset Doppler condition among the multiple Doppler velocity characteristics; if so, determine the feature of the target object that meets the preset Doppler condition as the first motion state feature of the target object; if not, determine the feature of the target object that does not meet the preset Doppler condition as the first motion state feature of the target object.

[0170] In one possible implementation, the feature determination module 803 is specifically configured to calculate the Doppler velocity feature between each track point in the track point interval using the following formula:

[0171]

[0172] Among them, th is the Doppler velocity characteristic between two track points in the track point interval, = , is the Doppler velocity of track point 1 within the track point interval, is the Doppler velocity of track point 2 within the track point interval, , , is the lateral movement speed of the target object, is the longitudinal motion speed of the target object, is the motion range angle, It is the difference between the azimuths of two track points in the track point interval.

[0173] In one possible implementation, the classification module 804 is specifically configured to determine an initial classification result corresponding to the target object based on the motion track point set; and determine a target category of the target object based on the initial classification result and the first motion state feature.

[0174] In one possible implementation, the classification module 804 is specifically used to obtain a first preset number of consecutive motion track frames corresponding to the target object; determine the target object size information and the number of motion track points corresponding to each motion track frame; determine whether the number of motion track frames whose target object size information meets a preset size threshold and whose number of motion track points is greater than a preset number of track points is greater than a second preset number; if not, determine the motion track confidence and motion distance corresponding to the target object; and determine an initial classification result corresponding to the target object based on the motion track confidence and the motion distance.

[0175] In one embodiment, the classification module 804 is specifically used to determine that the category of the target object is the first category if the target object size information meets a preset size threshold and the number of motion track frames is greater than the preset number of motion track frames and is greater than the second preset number, and use the first category as the initial classification result.

[0176] In one embodiment, the classification module 804 is also used to determine that the category of the target object is the second category if the movement distance meets a preset distance threshold and the movement track confidence is less than a first preset confidence threshold, and use the second category as an initial classification result; if the movement track confidence is greater than the second preset confidence threshold, determine that the category of the target object is the third category, and use the third category as an initial classification result; if the movement track confidence is greater than or equal to the first preset confidence threshold and less than or equal to the second preset confidence threshold, determine that the category of the target object is the fourth category, and use the fourth category as the initial classification result, and the first category, the second category, the third category and the fourth category are all different.

[0177] In one possible implementation, the classification module 804 is specifically used to obtain the movement speed of the target object based on the initial classification result; and determine the target category of the target object according to the first motion state characteristics, the motion speed and the motion track confidence.

[0178] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device and a readable storage medium.

[0179] Figure 8 A schematic block diagram of an example electronic device 800 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0180] like Figure 8 As shown, the device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 to a random access memory (RAM) 803. In the RAM 803, various programs and data required for the operation of the device 800 can also be stored. The computing unit 801, the ROM 802, and the RAM 803 are connected to each other via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0181] A number of components in the device 800 are connected to the I / O interface 805, including: an input unit 806, such as a keyboard, a mouse, etc.; an output unit 807, such as various types of displays, speakers, etc.; a storage unit 808, such as a disk, an optical disk, etc.; and a communication unit 809, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 809 allows the device 800 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0182] The computing unit 801 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 801 performs the various methods and processes described above, such as the target classification method. For example, in some embodiments, the target classification method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 800 via the ROM 802 and / or the communication unit 809. When the computer program is loaded into the RAM 803 and executed by the computing unit 801, one or more steps of the target classification method described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to perform the target classification method in any other appropriate manner (e.g., by means of firmware).

[0183] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0184] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0185] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0186] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0187] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0188] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0189] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0190] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present disclosure, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0191] The above is only a specific embodiment of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present disclosure, which should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A target classification method, characterized in that: The method comprises: Get the target object's motion track point set; Based on the set of motion track points, determining a motion interval angle corresponding to the target object; Determining a first motion state feature of the target object based on the motion interval angle and the motion track point set; The category of the target object is determined according to the first motion state feature.

2. The method according to claim 1, characterized in that The determining, based on the set of movement track points, a movement interval angle corresponding to the target object includes: The following formula is used to determine the motion interval angle corresponding to the target object based on the motion track point set: ； in, is the motion interval angle corresponding to the target object, R is the motion distance of the target object, and r is the maximum distance between the track points corresponding to the target object.

3. The method according to claim 1, characterized in that: The determining, based on the motion interval angle and the motion track point set, a first motion state feature of the target object comprises: Determine the track point interval corresponding to the movement interval angle; Calculating the Doppler velocity characteristics between each track point in the track point interval; Determining whether there is a Doppler velocity feature satisfying a preset Doppler condition among the plurality of Doppler velocity features; If so, determining the feature of the target object satisfying the preset Doppler condition as the first motion state feature of the target object; If not, the feature that the target object does not satisfy the preset Doppler condition is determined as the first motion state feature of the target object.

4. The method according to claim 3, characterized in that The calculating of the Doppler velocity characteristics between each track point in the track point interval includes: The Doppler velocity characteristics between each track point in the track point interval are calculated using the following formula: ； Among them, th is the Doppler velocity characteristic between two track points in the track point interval, = , is the Doppler velocity of track point 1 within the track point interval, is the Doppler velocity of track point 2 within the track point interval, , , is the lateral movement speed of the target object, is the longitudinal motion speed of the target object, is the motion range angle, It is the difference between the azimuths of two track points in the track point interval.

5. The method according to claim 4, characterized in that The determining the category of the target object according to the first motion state feature includes: Determine an initial classification result corresponding to the target object based on the motion track point set; A target category of the target object is determined based on the initial classification result and the first motion state feature.

6. The method according to claim 5, characterized in that The determining the initial classification result corresponding to the target object based on the moving track point set includes: Acquire a first preset number of continuous motion track frames corresponding to the target object; Determine the target object size information and the number of motion track points corresponding to each motion track frame; Determine whether the number of motion track frames in which the target object size information meets a preset size threshold and the number of motion track is greater than a preset number of track is greater than a second preset number; If not, determining the movement track confidence and movement distance corresponding to the target object; Based on the motion track confidence and the motion distance, an initial classification result corresponding to the target object is determined.

7. The method according to claim 6, characterized in that The method further comprises: If the target object size information meets the preset size threshold and the number of motion track frames with a motion point number greater than the preset point track number is greater than the second preset number, the category of the target object is determined to be the first category, and the first category is used as the initial classification result.

8. The method according to claim 6, characterized in that The determining, based on the motion track confidence and the motion distance, an initial classification result corresponding to the target object includes: If the movement distance satisfies a preset distance threshold and the movement track confidence is less than a first preset confidence threshold, determining that the category of the target object is a second category, and taking the second category as an initial classification result; If the motion track confidence is greater than a second preset confidence threshold, determining that the category of the target object is the third category, and using the third category as an initial classification result; If the motion track confidence is greater than or equal to the first preset confidence threshold and less than or equal to the second preset confidence threshold, the category of the target object is determined to be the fourth category, and the fourth category is used as the initial classification result. The first category, the second category, the third category and the fourth category are all different.

9. The method according to claim 8, characterized in that The determining the target category of the target object based on the initial classification result and the first motion state feature includes: Based on the initial classification result, obtaining the moving speed of the target object; The target category of the target object is determined according to the first motion state feature, the motion speed and the motion track confidence.

10. A target classification device, characterized in that: The device comprises: A track information acquisition module is used to obtain a set of moving track points of a target object; An interval angle determination module is used to determine the movement interval angle corresponding to the target object based on the movement track point set; A feature determination module, configured to determine a first motion state feature of the target object based on the motion interval angle and the motion track point set; A classification module is used to determine the category of the target object according to the first motion state feature.

Images