Target recognition method, device, electronic device and computer-readable storage medium

Abstract

Embodiments of the present application disclose a target recognition method, device, electronic device, and computer-readable storage medium. The method includes: obtaining first recognition information of at least one lane line in a road and second recognition information of at least one obstacle around a vehicle traveling on the road; fusing the first recognition information and the second recognition information to obtain relative position information between the lane line of the road and the obstacle around the vehicle; determining a target obstacle and a non-target obstacle from at least one obstacle according to the relative position information between the lane line of the road and the obstacle around the vehicle and the motion states of the obstacles within a preset range around the vehicle, thereby not only improving the efficiency of the ADAS in recognizing obstacles, but also improving the accuracy of the ADAS in recognizing obstacles.

Description

Technical Field

[0001] This application relates to the field of assisted driving technology, and particularly to a target recognition method, apparatus, electronic device, and computer-readable storage medium. Background Art

[0002] An Advanced Driving Assistance System (ADAS) utilizes various sensors installed on a vehicle, such as intelligent vision sensors, millimeter-wave radars, etc. It can sense the surrounding environment and collect data at any time during vehicle driving to identify, detect, and track static and dynamic objects, and combine navigation map data for system operation and analysis, thereby enabling the driver to perceive potential dangers in advance and effectively improving the comfort and safety of vehicle driving.

[0003] Among them, an intelligent vision sensor can provide high-resolution and high-clarity image information, and through deep learning algorithms, it can achieve the detection, tracking, and recognition of targets. However, in some special cases, the intelligent vision sensor may experience failures or reduced accuracy, such as low-light conditions, strong light interference, rain, snow, or occlusion, which will affect the perception effect of the intelligent vision sensor on targets. The millimeter-wave radar is not affected by factors such as light, weather, and occlusion, and it can stably provide information such as the distance, speed, and position of targets in different environments. However, the resolution of the millimeter-wave radar is relatively low, making it difficult to depict the details of targets, and it is very easy to identify some common targets on the road, such as street trees, road guardrails, and road signs, as obstacles around the vehicle.

[0004] Therefore, in the prior art, the data collected by the intelligent vision sensor is usually fused with the point cloud data collected by the millimeter-wave radar to improve the accuracy of obstacle recognition and classification. However, in some special cases, such as rain, snow, fog, dim light, glare, etc., the ADAS mainly relies on the perception of the millimeter-wave radar to identify obstacles, resulting in poor accuracy of obstacle recognition by the ADAS. Moreover, the data collected by the intelligent vision sensor and the millimeter-wave radar is relatively large, and the amount of calculation is large, resulting in low efficiency of obstacle recognition. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, this application provides a target recognition method, apparatus, electronic device, and computer-readable storage medium, aiming to solve the technical problems of low accuracy and efficiency in obstacle recognition of the advanced driving assistance system in the prior art.

[0006] To solve the above problems, an embodiment of this application provides a target recognition method, which includes:

[0007] Obtain first identification information of at least one lane line on a road and second identification information of at least one obstacle around a vehicle traveling on the road;

[0008] Fuse the first identification information and the second identification information to obtain relative position information between the lane lines of the road and the obstacles around the vehicle;

[0009] Determine a target obstacle and non-target obstacles from the at least one obstacle according to the relative position information between the lane lines of the road and the obstacles around the vehicle and the motion states of the obstacles within a preset range around the vehicle, where the motion states of the obstacles include a stationary state and a moving state.

[0010] In a second aspect, an embodiment of the present application further provides a target recognition device, which includes:

[0011] An obtaining unit, configured to obtain first identification information of at least one lane line on a road and second identification information of at least one obstacle around a vehicle traveling on the road;

[0012] A fusing unit, configured to fuse the first identification information and the second identification information to obtain relative position information between the lane lines of the road and the obstacles around the vehicle;

[0013] An identifying unit, configured to determine a target obstacle and non-target obstacles from the at least one obstacle according to the relative position information between the lane lines of the road and the obstacles around the vehicle and the motion states of the obstacles within a preset range around the vehicle, where the motion states of the obstacles include a stationary state and a moving state.

[0014] In a third aspect, an embodiment of the present application further provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where when the processor executes the computer program, the target recognition method described in the first aspect above is implemented.

[0015] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor is caused to execute the target recognition method described in the first aspect above.

[0016] The target recognition method, device, electronic device, and computer-readable storage medium provided by the embodiments of the present application fuse the first recognition information of the lane lines in the road and the second recognition information of the obstacles around the vehicle traveling on the road to obtain the relative position information between the lane lines and the obstacles around the vehicle. Furthermore, the target obstacles and non-target obstacles can be determined according to the relative position information between the obstacles and the lane lines and the motion state of the obstacles. The present application uses the position relationship between the lane lines and the obstacles for recognition, and combines the motion state of the obstacles to determine the target obstacles and non-target obstacles, with multiple consideration factors, improving the accuracy of ADAS in recognizing obstacles. At the same time, it does not require additional costs, reduces data fusion, and improves the efficiency of ADAS in recognizing obstacles. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 It is a schematic flowchart of the target recognition method provided by the embodiments of the present application;

[0019] Figure 2 It is a schematic flowchart of the target recognition method provided by the embodiments of the present application;

[0020] Figure 3 It is another schematic flowchart of the target recognition method provided by the embodiments of the present application;

[0021] Figure 4 It is a schematic diagram of the lane lines not recognized in the embodiments of the present application

[0022] Figure 5 It is a schematic diagram of the lane lines on both sides of the vehicle recognized in the embodiments of the present application;

[0023] Figure 6 It is a schematic diagram of the lane lines missing on both the left and right sides of the vehicle and the filling operation in the embodiments of the present application;

[0024] Figure 7 It is a schematic diagram of the lane lines missing on both the left and right sides of the vehicle and unable to be filled in the embodiments of the present application;

[0025] Figure 8 It is another schematic flowchart of the target recognition method provided by the embodiments of the present application;

[0026] Figure 9Another flowchart of the target recognition method provided by the embodiments of the present application;

[0027] Figure 10 Another flowchart of the target recognition method provided by the embodiments of the present application;

[0028] Figure 11 Schematic diagram of the distance between an obstacle and a lane line provided by the embodiments of the present application;

[0029] Figure 12 Schematic diagram of the edge line of a carriageway on a road provided by the embodiments of the present application;

[0030] Figure 13 Schematic diagram of a road guardrail and road signs on a road provided by the embodiments of the present application;

[0031] Figure 14 Another flowchart of the target recognition method provided by the embodiments of the present application;

[0032] Figure 15 Another flowchart of the target recognition method provided by the embodiments of the present application;

[0033] Figure 16 Another flowchart of the target recognition method provided by the embodiments of the present application;

[0034] Figure 17 Another flowchart of the target recognition method provided by the embodiments of the present application;

[0035] Figure 18 Schematic block diagram of the target recognition device provided by the embodiments of the present application;

[0036] Figure 19 Schematic block diagram of the electronic device provided by the embodiments of the present application. Detailed implementation manners

[0037] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0038] It should be understood that when used in this specification and the appended claims, the terms "comprises" and "comprising" indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their groups.

[0039] It should also be understood that the terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. As used in the specification of this application and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0040] It should be further understood that the term "and / or" used in the specification of this application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0041] The target recognition method provided by the embodiments of this application is applied to the in-vehicle terminal device where the ADAS is located. This method is executed by the application software installed in the in-vehicle terminal device where the ADAS is located to perform the target recognition method.

[0042] It should be noted that the application scenarios of the above embodiments are only examples. The services and scenarios described in the embodiments of this application are for more clearly explaining the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those of ordinary skill in the art can know that with the evolution of the system and the emergence of new service scenarios, the technical solutions provided by the embodiments of this application are equally applicable to similar technical problems. The following will be described in detail respectively.

[0043] It should also be noted that the description order of the following embodiments does not serve as a limitation on the preferred order of the embodiments. The following will describe the target recognition method in detail.

[0044] The following will describe the target recognition method in detail. Please refer to Figure 1 , Figure 1 which is the flowchart of the target recognition method provided by the embodiments of this application.

[0045] As Figure 1 shown, this method includes the following steps S110 to S130.

[0046] S110. Obtain the first recognition information of at least one lane line in the road and the second recognition information of at least one obstacle around the vehicle traveling on the road.

[0047] In this embodiment, the first recognition information is the characteristic information of the lane line in the road, and the lane line in the road can be recognized through the first recognition information; the second recognition information is the characteristic information of the obstacle around the vehicle traveling on the road, and the obstacle during the vehicle traveling on the road can be recognized through the second recognition information. Both the first recognition information and the second recognition information can be obtained from the data collected by various sensors in the ADAS.

[0048] Among them, the lane lines on the road are used as one of the reference objects for screening a plurality of identified obstacles to determine the target obstacle in this application. That is, after obtaining the lane lines on the road and a plurality of obstacles around the vehicle, the lane lines are used as a reference object to screen the obstacles, so as to determine the target obstacle and non-target obstacle among the plurality of obstacles.

[0049] Currently, when a vehicle is driving on the road, it usually encounters harsh conditions such as low light conditions, strong light interference, rain, snow, fog, and occlusion, resulting in poor perception effects of the targets of intelligent vision sensors. However, the materials generally used for lane lines on the road are solvent-based coatings, which have strong reflectivity. At the same time, the lane lines on the road are usually relatively close to the vehicle. Furthermore, it can enable the vision sensors in ADAS to accurately collect the lane lines on the road without worrying about inaccurate collection of lane lines by intelligent vision sensors due to harsh conditions such as low light conditions, strong light interference, rain, snow, fog, and occlusion. Therefore, this application can accurately obtain the first recognition information of the lane lines on the road directly from the image information collected by the vision sensors in ADAS.

[0050] At the same time, the radar in ADAS can stably provide information such as the distance, speed, and position of the target in different environments. Therefore, this application preferentially obtains the second recognition information of the obstacles around the vehicle from the data collected by the radar in ADAS, and then fuses the first recognition information and the second recognition information. Furthermore, it can greatly reduce the fusion of the data collected by various sensors, thereby improving the efficiency and accuracy of identifying the target obstacles around the vehicle.

[0051] In other invention embodiments, as Figure 2 shown, step S110 includes steps S111 and S112.

[0052] S111. Obtain the first recognition information of at least one lane line on the road according to the preset vision sensor;

[0053] S112. Obtain the second recognition information of at least one obstacle around the vehicle driving on the road according to the preset radar.

[0054] In this embodiment, the vision sensor and the radar are both sensors in ADAS on the vehicle. The radar can be a millimeter-wave radar, a lidar, etc. This application preferentially selects a millimeter-wave radar, which can stably detect obstacles and is less affected by extreme weather, and roughly distinguishes the types of obstacles according to the RCS (Radar Cross-Section). (For pedestrians, it is about 1m 2 , and for small cars, it is about 10m 2)。The detection range of a millimeter-wave radar is generally obstacles within 250 meters, and it can measure the relative speed of a target based on the Doppler effect. However, its detection accuracy is lower than that of a lidar. Due to its working principle, a millimeter-wave radar is prone to identifying noise targets such as roadside trees, metal street lights, metal road guardrails, metal road signs, and metal trash cans as vehicles or pedestrians.

[0055] In an embodiment of the present application, first recognition information of lane lines in a road is extracted based on a vision sensor in ADAS, and second recognition information of several obstacles around a vehicle in the road is extracted from data collected by a radar in ADAS. Then, the first recognition information and the second recognition information are fused, so that the data calculation amount for identifying target obstacles around the vehicle can be greatly reduced, and at the same time, target obstacles around the vehicle, such as vehicles and pedestrians, can be accurately identified under harsh conditions such as low light conditions, strong light interference, rain, snow, fog, and occlusion.

[0056] It should be noted that both the first recognition information and the second recognition information can be obtained from data collected by various sensors in the vehicle's built-in ADAS, or can be obtained from a recognition module subsequently configured by the vehicle user. The acquisition methods of the first recognition information and the second recognition information can be selected according to actual applications, and the present application does not make specific limitations.

[0057] In other embodiments of the invention, such as Figure 3 shown, step S111 includes steps S1111, S1112, S1113, and S1114.

[0058] S1111. Obtain initial lane line recognition information including at least one lane line;

[0059] S1112. Based on the initial lane line recognition information, determine whether lane lines are recognized on the left and right sides of the vehicle. If at least one lane line is recognized on both the left and right sides of the vehicle, use the initial lane line recognition information as the first recognition information; otherwise, execute step S1113;

[0060] S1113. Based on the initial lane line recognition information, determine whether a lane line filling operation can be performed. If a lane line filling operation can be performed, perform a lane line filling operation on the initial lane line recognition information to obtain filled recognition information, and use the filled recognition information as the first recognition information; otherwise, use the initial lane line recognition information as the first recognition information.

[0061] Since the lane line can be a straight line or a curve, in the coordinate system, the lane line can be described as a line on a two-dimensional plane, which can be characterized by one or more functional expressions. The obstacles on the road can be described by the coordinates of a certain point in the same coordinate system as the lane line. Among them, the first position information of the lane line can be described by a cubic function curve as follows: x = a + by + cy2 + dy3, where y ∈ [y min , y max . Where x is the horizontal axis with the right direction as the positive direction; y is the vertical axis with the front direction as the positive direction; a, b, c, d, y min and y max can be six parameters given by the vision sensor. a is the offset between the lane line and the vehicle coordinate system, b is the lane line yaw angle, c is the lane line curvature, d is the lane line curvature change rate, y min = 0 and y max = 0 represent the perception results without vision.

[0062] In bad weather such as rain and snow, the perception effect of the vision sensor is poor and it often cannot recognize the complete lane line. In other embodiments, considering that the lane line recognition results of adjacent frames may mutate, therefore, step S1111 further includes: correcting the lane line recognition result of the current frame based on the lane line recognition result of the previous frame, as follows:

[0063] Obtain the third recognition information of the lane line on the road at the current moment, and obtain the fourth recognition information of the lane line on the road at the previous moment. Among them, the third recognition information of the lane line on the road at the current moment is determined based on the captured image of the vision sensor at the current moment, and the fourth recognition information of the lane line on the road at the previous moment is determined based on the captured image of the vision sensor at the previous moment;

[0064] If the number of lane lines in the third recognition information is less than the number of lane lines in the fourth recognition information, perform the lane line correction operation; otherwise, do not perform the lane line correction operation, and use the third recognition information as the initial lane line recognition information;

[0065] The lane line correction operation includes: if the third recognition information does not include the recognition information of at least one lane line, use the fourth recognition information as the initial lane line recognition information; otherwise, match each lane line in the third recognition information among the multiple lane lines in the fourth recognition information, add at least one lane line in the fourth recognition information that does not match the lane line in the third recognition information to the third recognition information, and use the third recognition information with the lane lines in the fourth recognition information added as the initial lane line recognition information.

[0066] Among them, the matching of the lane lines in the third recognition information with the lane lines in the fourth recognition information is as follows:

[0067] For the same lane line, the curves described by the lane line data of adjacent frames should not have relatively abnormal differences. Therefore, at least one lane line corresponding to the lane line in the third recognition information will be found from multiple lane lines in the fourth recognition information according to the similarity between the coefficients of the lane line curves.

[0068] By introducing the lane lines of the previous frame to correct the lane lines of the current frame, the number of lane lines can be increased to a certain extent, the vehicle driving range formed by the lane lines can be expanded, and the reference for filtering and screening obstacles is carried out in a larger vehicle driving range, ensuring driving safety.

[0069] Specifically, in the embodiment of the present application, lane line recognition is performed based on the collected images of the vision sensor. The collected images are the images collected by the vision sensor on the road. After obtaining the collected images, the feature information of the lane lines on the road can be extracted from the collected images according to the combination of deep learning algorithms. The feature information includes the position information and attitude information of the lane lines. During the driving process of the vehicle on the road, not all lane lines of the road exist in the images collected by the vision sensor. Especially when the vehicle is driving in a harsh environment, such as when the vision is poor, the images collected by the vision sensor usually lack some lane lines of the road. Since the present application performs the recognition of target obstacles and non-target obstacles with the lane lines as the reference, after obtaining the images collected by the vision sensor, the present application extracts the lane lines to obtain the initial lane line recognition information and checks the initial lane line recognition information. If no lane lines are recognized in the collected images, as Figure 4 shown, that is, there are no lane lines in the initial lane line recognition information, it means that there are no lane lines as the reference. Therefore, the first recognition information is not obtained, the target recognition is ended, and the collected images at the next moment are waited for, and the lane line recognition is performed again to obtain the initial lane line recognition information. It should be noted that the end of the target recognition here refers to the end of the method for target recognition provided by the present application based on the first recognition information and the second recognition information. In fact, other target recognition methods can be integrated in the ADAS and executed when the target recognition method provided by the present application ends. For example, the second recognition information of each obstacle around the vehicle can be directly used as the target recognition result, and each obstacle is a target obstacle, so as to avoid no target recognition result being output, thereby facilitating the ADAS to perform driving control.

[0070] The target recognition logic of the embodiments of this application is to determine non-target obstacles and target obstacles. Among them, the judgment principle of non-target obstacles is as follows: Obstacles located outside the driving range that the vehicle can recognize and far away from the driving range that the vehicle can recognize are non-target obstacles, and stationary obstacles located outside the driving range that the vehicle can recognize and close to the driving range that the vehicle can recognize are non-target obstacles. Therefore, it is necessary to determine the driving range that the vehicle can recognize.

[0071] The reference for non-target obstacles is the driving range that the vehicle can recognize. The driving range is determined by the lane lines on both sides of the vehicle. Therefore, it is necessary to check whether lane lines are recognized on the left and right sides of the vehicle. If at least one lane line is recognized on both the left and right sides of the vehicle, as Figure 5 in Figure 5 a, Figure 5 b, Figure 5 c, Figure 5 d shown, then the leftmost lane line farthest from the vehicle can be found among at least one lane line on the left side of the vehicle, and the rightmost lane line farthest from the vehicle can be found among at least one lane line on the right side of the vehicle. Based on the leftmost lane line and the rightmost lane line, the driving range that the vehicle can recognize can be determined, so as to recognize non-target obstacles.

[0072] If the lane lines are missing on both the left and right sides of the vehicle, as Figure 6 , Figure 7 shown, then it is necessary to judge whether the lane line filling operation can be performed. The purpose of the filling operation is to make sure that there is at least one lane line on both the left and right sides of the vehicle. If the filling operation cannot be performed, the driving range of the vehicle cannot be determined, so the subsequent recognition of non-target obstacles is not carried out.

[0073] Among them, in the process of judging whether the lane line filling operation can be performed and filling the lane lines, it includes:

[0074] Judging the missing state of the lane lines. The missing state includes: the lane lines are missing on one side of the vehicle and the lane lines are missing on both sides of the vehicle; among them, the lane lines are missing on both sides of the vehicle, and at least one lane line is included in the initial recognition information, indicating that the recognized lane lines are in the middle of the vehicle.

[0075] For the missing state where lane lines are missing on one side of the vehicle, translate the lane line closest to the vehicle's central axis by a distance H towards the side where the lane lines are missing from the vehicle, to obtain a virtual lane line. If the vehicle is within the lane range formed by the lane line closest to the vehicle and the virtual lane line, lane line filling operation can be performed, and the position information and attitude information of the virtual lane line are added to the initial recognition information to obtain filled recognition information. If the vehicle is not within the lane formed by the lane line closest to the vehicle and the virtual lane line, lane line filling operation cannot be performed, where H is the set lane width;

[0076] For the missing state where lane lines are missing on both sides of the vehicle, translate the lane line in the middle of the vehicle towards the left and right sides by a distance H respectively, to obtain two virtual lane lines. If the vehicle is within the lane range formed by the two virtual lane lines, lane line filling operation can be performed, and the position information and attitude information of the virtual lane line are added to the initial recognition information to obtain filled recognition information. If the vehicle is not within the lane range formed by the two virtual lane lines, lane line filling operation cannot be performed, where H is the set lane width.

[0077] In this embodiment, the value of H is the lane width under the urban road design standard. For example, it is 3.5m. In other embodiments, if the initial recognition information includes at least two lane lines, the lane width is calculated based on the distance between two adjacent lane lines among the at least two lane lines, and the calculated lane width is used as the value of H. For example, if two lane lines are recognized on one side of the vehicle, the distance between these two lane lines is the lane width.

[0078] It should be noted that the lane width calculated from the distance between two adjacent lane lines may be the width of one or more lanes. For example, there may be an unrecognized lane line between the two recognized lane lines. In this case, the distance between the two adjacent recognized lane lines is actually the width of two lanes. In a feasible implementation, a lane width threshold is set. When the calculated lane width is less than the lane width threshold, the calculated lane width is used as the value of H. When the calculated lane width is greater than the lane width threshold, the lane width under the urban road design standard is used as the value of H, where the value of the lane width threshold is based on the urban road design standard. For example, the lane width threshold is taken as 3.8m.

[0079] As Figure 6 shown, Figure 6 a, Figure 6 b, Figure 6 c show the schematic of the lane line filling operation that can be performed in the missing state where lane lines are missing on one side of the vehicle, Figure 6d shows the operation of lane line filling when there is a missing state of lane lines on both sides of the vehicle. As Figure 7 shown, Figure 7 a, Figure 7 b, Figure 7 c, Figure 7 d shows the schematic of judging that lane line filling operation cannot be performed in the missing state of lane lines on one side of the vehicle. The main purpose of this application is to identify the targets detected by the radar with the reference of lane lines and find out the obstacles misdetected among them, that is, non-target obstacles. And the second identification information of the obstacles and the first identification information of the lane lines are not in the same coordinate system, so it is necessary to fuse the first identification information and the second identification information.

[0080] S120. Fuse the first identification information and the second identification information to obtain the relative position information between the lane lines of the road and the obstacles around the vehicle.

[0081] In this embodiment, the relative position information is used to characterize the positional relationship between the lane lines and the obstacles. When the first identification information is fused with the second identification information, the first identification information and the second identification information can be directly mapped into the same coordinate system, and then the relative position information between the lane lines of the road and the obstacles around the vehicle can be calculated. Among them, the coordinate system can be a certain coordinate system of the vision sensor or a pre-constructed coordinate system.

[0082] In other embodiments of the invention, as Figure 8 shown, step S120 includes steps S1201 and S1202.

[0083] S1201. Based on a preset coordinate system, fuse the first identification information and the second identification information to obtain the first coordinate information of the lane lines of the road and the second coordinate information of the obstacles around the vehicle;

[0084] S1202. Obtain the relative position information between the lane lines of the road and the obstacles around the vehicle according to the first coordinate information and the second coordinate information.

[0085] In this embodiment, the preset coordinate system is a certain coordinate system of the vision sensor. At this time, the second identification information of the obstacles can be directly mapped into this coordinate system to be fused with the first identification information, and then the first coordinate information of the lane lines of the road and the second coordinate information of the obstacles around the vehicle can be obtained respectively. Then, the relative position information between the lane lines of the road and the obstacles around the vehicle can be generated through the first coordinate information and the second coordinate information.

[0086] This application is mainly to classify the obstacles around the vehicle recognized by the radar into target obstacles and non-target obstacles. Among them, target obstacles are those that will affect the vehicle's driving, such as vehicles and pedestrians in the vehicle's lane. Non-target obstacles are those that will not affect the vehicle's driving, such as vehicles and pedestrians far from the vehicle's lane, trees and street lights on both sides of the road, etc. Therefore, on the one hand, the lane lines need to be used as a reference to determine the position of the obstacles relative to the vehicle's lane, and on the other hand, the motion states of the obstacles need to be considered.

[0087] S130. Determine target obstacles and non-target obstacles from the at least one obstacle according to the relative position information between the lane lines of the road and the obstacles around the vehicle and the motion states of the obstacles within a preset range around the vehicle, where the motion states of the obstacles include a stationary state and a moving state.

[0088] In other embodiments of the invention, as Figure 9 shown, step S130 includes steps S1301, S1302, and S1303.

[0089] S1301. Determine the leftmost lane line and the rightmost lane line based on the first recognition information, where the lane line in the first recognition information that is on the left side of the vehicle and the farthest from the vehicle is the leftmost lane line, and the lane line in the first recognition information that is on the right side of the vehicle and the farthest from the vehicle is the rightmost lane line;

[0090] S1302. Determine non-target obstacles from the at least one obstacle, where the non-target obstacles include: obstacles located on the left side of the leftmost lane line and with a distance greater than a first preset distance from the leftmost lane line, obstacles located on the right side of the rightmost lane line and with a distance greater than a first preset distance from the rightmost lane line, obstacles with a stationary state and located on the left side of the leftmost lane line and with a distance less than the first preset distance and greater than a second preset distance from the leftmost lane line, and obstacles with a stationary state and located on the right side of the rightmost lane line and with a distance less than the first preset distance and greater than a second preset distance from the rightmost lane line;

[0091] S1303. Based on the non-target obstacles, determine target obstacles and non-target obstacles from the at least one obstacle;

[0092] Among them, in step S130, the obstacles within the preset range around the vehicle include: obstacles located on the left side of the leftmost lane line and with a distance less than the first preset distance and greater than a second preset distance from the leftmost lane line, and obstacles located on the right side of the rightmost lane line and with a distance less than the first preset distance and greater than a second preset distance from the rightmost lane line.

[0093] Specifically, the lane line is represented as a curve in the coordinate system, and the obstacle is represented as a point or a set of multiple points in the coordinate system. The calculation method of the distance between the obstacle and the lane line is the distance from the center point of the obstacle to the lane line curve, that is, the distance from a point to a line. Generally, the calculation means is to calculate the vertical distance (i.e., the shortest distance from a point to a line). In other embodiments of the present invention, in order to simplify the calculation, the lateral projection distance from the point to the line is used as the distance between the obstacle and the lane line. For example, as Figure 11 shown, the center point P of the obstacle is laterally projected onto the line L, and the projection distance is D, and D is the distance between the obstacle and the lane line.

[0094] Specifically, the non-target obstacles are divided into two categories. The first category of non-target obstacles are obstacles that are not within the driving range and are far from the driving range (i.e., outside the outermost lane line and far from the outermost lane line) by a first preset distance. The second category of non-target obstacles are stationary obstacles that are not within the driving range and are close to the driving range (i.e., outside the outermost lane line and close to the outermost lane line). The distance between the second non-target obstacle and the outermost lane line is less than the first preset distance and greater than the second preset distance.

[0095] Under normal circumstances, the vehicle travels in the lane, and the traveling direction of the vehicle is also within the lane. Obstacles that are far from the lane will not affect the driving control of the vehicle. At this time, the obstacles far from the lane can be filtered without missing the obstacles that may cause danger. When the vehicle completely identifies all lane lines, the leftmost lane line and the rightmost lane line are the edge lane lines of the motor vehicle lane where the vehicle is located. At this time, the obstacles outside the leftmost lane line and the rightmost lane line are actually obstacles outside the motor vehicle lane. According to the industry standard CJJ 37-2012 "Urban Road Engineering Design Specification", the width of the non-motor vehicle lane combined with the motor vehicle should not be less than 2.5 m. Therefore, the obstacles located to the left of the leftmost lane line and more than 2.5 m away from the leftmost lane line (i.e., the first category of non-target obstacles) are regarded as non-target obstacles, and the obstacles located to the right of the rightmost lane line and more than 2.5 m away from the rightmost lane line (i.e., the first category of non-target obstacles) are regarded as non-target obstacles, that is, the obstacles outside the non-motor vehicle lane. Furthermore, a large number of noise targets can be filtered by filtering non-target obstacles. That is, the first preset distance can be 2.5 m. It can be understood that when the vehicle does not completely identify all lane lines, the leftmost lane line and the rightmost lane line obtained based on the first recognition information are not the edge lane lines of the motor vehicle lane. At this time, the first category of non-target obstacles are still identified based on the first preset distance and the leftmost lane line and the rightmost lane line. At this time, the first preset distance is actually to identify the obstacles far from the vehicle driving range as non-target obstacles.

[0096] In addition, according to the national standard GB 50688-2011 "Code for Design of Urban Road Traffic Facilities", street trees and street lamps are generally set outside the scope of the motor vehicle lane. Considering that street trees and street lamps are stationary relative to the ground, stationary obstacles outside the vehicle driving range and close to the driving range can be identified as the second type of non-target obstacles. Considering the urban road design specifications of the non-motor vehicle lane, and that being close to the driving range actually means being close to the outermost lane line, stationary obstacles (i.e., the second type of non-target obstacles) that are always stationary and located to the left of the leftmost lane line and have a distance from the leftmost lane line less than 2.5 m and greater than 30 cm are regarded as non-target obstacles, and stationary obstacles (i.e., the second type of non-target obstacles) that are always stationary and located to the right of the rightmost lane line and have a distance from the rightmost lane line less than 2.5 m and greater than 30 cm are regarded as non-target obstacles, that is, the stationary obstacles between the motor vehicle lane and the non-motor vehicle lane. Furthermore, a large number of noise targets can be filtered by filtering non-target obstacles. That is, the first preset distance can be 2.5 m, and the second preset distance is 30 cm. It can be understood that the lane line itself has a certain width, and the lane line feature information in the first recognition information is a curve. Therefore, twice the line width is taken as the judgment basis outside the lane line, that is, 30 cm away from the lane line. As Figure 12 shown, by identifying the second type of non-target obstacles, obstacles such as trees and street lamps near the edge lines of the carriageways on both sides of the road can be identified.

[0097] In other embodiments of the present invention, as Figure 10 shown, step S130 includes steps S1301', S1302', S1303', S1304', S1305';

[0098] S1301', determining the leftmost lane line and the rightmost lane line based on the first recognition information, wherein the lane line in the first recognition information that is located on the left side of the vehicle and the farthest from the vehicle is the leftmost lane line, and the lane line in the first recognition information that is located on the right side of the vehicle and the farthest from the vehicle is the rightmost lane line;

[0099] S1302', determining non-target obstacles from the at least one obstacle, wherein the non-target obstacles include: obstacles located to the left of the leftmost lane line and having a distance from the leftmost lane line greater than the first preset distance, obstacles located to the right of the rightmost lane line and having a distance from the rightmost lane line greater than the first preset distance, obstacles with a stationary motion state that are located to the left of the leftmost lane line and have a distance from the leftmost lane line less than the first preset distance and greater than the second preset distance, and obstacles with a stationary motion state that are located to the right of the rightmost lane line and have a distance from the rightmost lane line less than the first preset distance and greater than the second preset distance;

[0100] S1303', determine the obstacle to be corrected from the at least one obstacle, where the obstacle to be corrected includes: an obstacle located on the lane line of the road and in a stationary state;

[0101] S1304', correct the obstacle to be corrected into a target obstacle or a non-target obstacle;

[0102] S1305', determine the target obstacle and the non-target obstacle from the at least one obstacle based on the non-target obstacle, the corrected non-target obstacle, and the corrected target obstacle;

[0103] Among them, in step S130, the obstacles within the preset range around the vehicle include: an obstacle located on the left side of the leftmost lane line and having a distance from the leftmost lane line less than the first preset distance and greater than the second preset distance, an obstacle located on the right side of the rightmost lane line and having a distance from the rightmost lane line less than the first preset distance and greater than the second preset distance, and an obstacle located on the lane line of the road.

[0104] In this embodiment, the concept of the obstacle to be corrected is introduced. The judgment principle of the obstacle to be corrected is: a stationary obstacle located on the lane line. The obstacles located on the lane line may be vehicles, pedestrians, road signs, road guardrails, etc. Among them, the moving obstacles will obviously affect the vehicle driving and need to be concerned. While the stationary obstacles may be road signs, road guardrails, stationary vehicles, stationary pedestrians, etc., and it is necessary to determine whether to pay attention according to the specific situation. Since the radar has low detection accuracy, it may misidentify the target.

[0105] Specifically, since there will be obstacles such as Figure 13 shown road signs and road guardrails on the oncoming lane lines of the road. Such obstacles are stationary obstacles, and such obstacles may be misdetected by the radar, resulting in classifying such obstacles as the same level of obstacles as the pedestrians and vehicles around the vehicle. Among them, the oncoming lane line is a line used to divide two reverse lanes, which can be a solid line, a dotted line or a double solid line, etc. The edge line of the carriageway is a line used to indicate the edge of the motor vehicle lane or to divide the boundary between the motor vehicle and the non-motor vehicle lane, which can be a solid edge line, a dotted edge line, etc.

[0106] Therefore, after obtaining the relative position information between at least one lane line of the road and the obstacle, the present application identifies the obstacle located on the lane line of the road from the relative position information, and then judges the motion state of the obstacle located on the lane line, and identifies the stationary obstacle as the obstacle to be corrected.

[0107] Among them, according to the national standard GB 50688-2011 "Code for Design of Urban Road Traffic Facilities", road guardrails and road signs are generally set on or outside the lane lines for oncoming traffic, which can be the center double yellow line, single yellow solid line, center dotted line, lane dividing line, lane edge line, etc., and the width of the markings is generally 15 cm. Therefore, when identifying a stationary obstacle located on the lane line of a road, the judgment basis is a preset distance (such as 30 cm) from the lane line and a stationary state of motion.

[0108] It should be noted that since the vision sensor can also recognize the line width, line type, etc. of the lane line, the preset distance of 30 cm can also vary with the recognition result. Similarly, the second preset distance for identifying the second type of non-target obstacle can also vary with the recognition result.

[0109] According to the above analysis, it can be seen that the obstacles to be corrected include pedestrians, vehicles, and road signs and road guardrails that are misdetected as pedestrians and lanes. Therefore, this application corrects the obstacles to be identified according to the size of the obstacles.

[0110] As Figure 14 shown, correcting the obstacle to be corrected into a target obstacle or a non-target obstacle includes the following steps:

[0111] S13041. Obtain the radar cross section of the obstacle to be corrected from the second recognition information;

[0112] S13042. Correct the obstacle to be corrected into a target obstacle or a non-target obstacle according to the size relationship between the radar cross section of the obstacle to be corrected and the preset cross section threshold.

[0113] Specifically, this application obtains the radar cross section of the obstacle from the second recognition information of the obstacle, and then corrects this type of obstacle by judging whether the radar cross section is greater than the preset cross section threshold.

[0114] In addition, when judging whether the radar cross section is greater than the preset cross section threshold to correct this type of obstacle, the obstacle to be corrected with a radar cross section smaller than the preset cross section threshold (such as 5 m 2 ) is corrected, while the obstacle to be corrected with a radar cross section greater than the preset cross section threshold is used as a target obstacle. It should also be noted that when identifying a target obstacle, the corrected obstacle can be selected to be removed (i.e., corrected to a non-target obstacle), or it can be selected to be retained (i.e., corrected to a target obstacle). Whether this type of obstacle is removed can be selected according to the actual application, and this application does not make specific limitations.

[0115] Through the recognition of obstacles to be corrected and the correction of obstacles to be corrected, obstacles on the lane line and stationary obstacles can be recognized. According to the size of the radar cross section (RCS) of the obstacle, the obstacle to be corrected is corrected into a target obstacle and a non-target obstacle.

[0116] It should also be noted that in other embodiments, the concept of an obstacle to be corrected may not be introduced. For the obstacles on the lane line, according to their motion state and radar cross section, it is determined whether they are target obstacles or non-target obstacles. In this embodiment, when determining target obstacles and non-target obstacles, it is necessary to combine the position of the obstacle relative to the lane line, the motion state of the obstacle, and the radar cross section as the judgment basis. After introducing the obstacle to be corrected, only the radar cross section needs to be introduced as the judgment basis for the obstacles on the lane line, reducing the resource occupancy.

[0117] It can be understood that when recognizing obstacles to be corrected, the leftmost lane line and the rightmost lane line do not need to be considered. Therefore, when the first recognition information includes at least one lane line on the left side of the vehicle and at least one lane line on the right side of the vehicle, non-target obstacles and obstacles to be corrected can be determined. Then, the obstacle to be corrected is corrected, and the non-target obstacle obtained by the correction is recorded as the third type of non-target obstacle. Thus, the first type of non-target obstacle, the second type of non-target obstacle, and the third type of non-target obstacle, that is, all non-target obstacles, and the obstacles other than non-target obstacles are target obstacles. If there is no lane line on the left side of the vehicle or on the right side of the vehicle in the first recognition information, the recognition of the first type of non-target obstacle and the second type of non-target obstacle cannot be performed, and only the recognition of the obstacle to be corrected is performed. Thus, the third type of non-target obstacle is obtained by correcting the obstacle to be corrected.

[0118] Only determining the first type of non-target obstacle and the second type of non-target obstacle actually classifies the obstacles outside the vehicle driving range, and the obstacles within the vehicle driving range are all considered as target obstacles. This will cause some unconcerned obstacles to be considered as target obstacles, affecting the decision-making of the ADAS system. For this reason, this embodiment considers the obstacle to be corrected and judges the obstacles on the lane line, so as to further determine the non-target obstacles that do not need to be concerned.

[0119] In the above recognition of non-target obstacles and obstacles to be corrected, the motion state of the obstacle is considered. In other embodiments of the invention, as Figure 15 shown, the motion state of the obstacle is obtained based on the second recognition information. Determining the motion state of an obstacle includes step S130a and step S130b:

[0120] S130a. Obtain the instantaneous speed of the obstacle from the second identification information;

[0121] S130b. Obtain the motion state of the obstacle based on the magnitude relationship between the instantaneous speed of the obstacle and a preset speed threshold, where the motion state includes a stationary state and a moving state.

[0122] Since the data measured by the sensor usually has errors, an actually stationary obstacle may also have a measured motion speed by a sensor such as a radar. Therefore, when determining the motion state of the obstacle, it is necessary to first obtain the instantaneous speed of each obstacle from the point cloud information, and then determine whether the instantaneous speed of each obstacle is lower than the preset speed threshold. If it is lower than the preset speed threshold, the motion state of the obstacle is the stationary state; otherwise, the motion state of the obstacle is the moving state. Among them, the second identification information includes the instantaneous speed of the obstacle, and the second identification information can be obtained from the point cloud collected by the millimeter-wave radar.

[0123] Among them, for a 77GHz radar, its range resolution is about 5cm. Therefore, in practice, taking five times this value, i.e., 0.25m / s, as the determination speed for stationary state is sufficient, that is, the preset speed can be 0.25m / s.

[0124] In other embodiments of the invention, as Figure 16 shown, the motion state of the obstacle is obtained based on the second identification information. Determining the motion state of an obstacle includes steps S130a' and S130b'.

[0125] S130a'. Obtain the accumulated displacement of the obstacle within a preset time from the second identification information;

[0126] S130b'. Obtain the motion state of the obstacle based on the magnitude relationship between the accumulated displacement of the obstacle and a preset displacement threshold, where the motion state includes a stationary state and a moving state.

[0127] In this embodiment, there may also be pedestrians walking slowly among the target obstacles. When identifying stationary obstacles around the vehicle only by the instantaneous speed, the pedestrians walking slowly may be regarded as stationary obstacles. For this reason, the present application can also identify the motion state of the obstacle by obtaining the accumulated displacement of the obstacle within a preset time.

[0128] Specifically, when the accumulated displacement of the obstacle is lower than the preset displacement threshold, it can be determined that the obstacle is a stationary obstacle; if it is not lower than the preset displacement threshold, the obstacle is an obstacle in a motion state.

[0129] Among them, when obtaining the cumulative displacement of the obstacle, the vehicle's speedometer and inertial measurement unit can also be used to integrate the displacement of the vehicle itself, and then the displacement of the vehicle itself is used to correct the target position recognized by the millimeter-wave radar, and then the cumulative displacement of the obstacle can be obtained.

[0130] In addition, due to the measurement error of the sensor and the cumulative error of the calculation, a certain tolerance is also required for the cumulative displacement of the obstacle. In practice, the larger value of a hundred times the radar range resolution and 5% of the target distance can be taken as the determination basis for the cumulative displacement of the obstacle.

[0131] It should be noted that the present application can also identify stationary obstacles through instantaneous speed and cumulative displacement at the same time. When using instantaneous speed and cumulative displacement to identify stationary obstacles, the two identification steps can be interchanged, and the specific identification method can be selected according to actual applications. The present application does not make specific limitations.

[0132] In addition, as Figure 17 shown, in other embodiments of the present invention, the target recognition method further includes step S140, as follows:

[0133] S140. Filter at least one obstacle around the vehicle according to the non-target obstacle.

[0134] Filter out the non-target obstacles and retain the target obstacles, so as to only focus on the target obstacles and perform driving control, and the obstacles misdetected by the radar will not affect the subsequent processing.

[0135] In the embodiments of the present application, first obtain the position information of the obstacle relative to the lane line, and then examine the motion state and RCS size of the obstacles in a specific position range (obstacles located on the lane line, obstacles located on the left side of the leftmost lane line and less than the first preset distance and greater than the second preset distance from the leftmost lane line, obstacles located on the right side of the rightmost lane line and less than the first preset distance and greater than the second preset distance from the rightmost lane line). In other embodiments of the present invention, it can be understood that the motion states of each obstacle around the vehicle can also be examined first to obtain the first stationary obstacle in the stationary state and the second moving obstacle in the moving state. The first stationary obstacle is fused with the lane line to obtain the obstacle to be corrected and the second type of non-target obstacle, and the second moving obstacle is fused with the lane line to obtain the first type of non-target obstacle.

[0136] When the present application is applied to ADAS, obtain the first recognition information of the lane line and the second recognition information of the obstacle, obtain the target obstacle and the non-target obstacle, and then the non-target obstacles among all the obstacles can be filtered out and the target obstacles can be retained, which can filter and screen the targets recognized by the radar, and can filter out the misdetected obstacles and the obstacles that are not concerned about.

[0137] In the target recognition method provided by the embodiments of the present application, according to the relative positions of obstacles and lane lines, as well as the motion states and radar cross-sectional areas of the obstacles, the obstacles around the vehicle are divided into target obstacles and non-target obstacles. The non-target obstacles can be filtered out, so as to only focus on the target obstacles and perform driving control. This can not only improve the efficiency of ADAS in recognizing obstacles, but also improve the accuracy of ADAS in recognizing obstacles. Further, the present application directly obtains the first recognition information and the second recognition information through the intelligent vision sensor and radar in the ADAS respectively, without incurring additional costs, and can improve the accuracy and efficiency of obstacle recognition of the advanced driving assistance system.

[0138] The embodiments of the present application further provide a target recognition device 100, which is used to execute any one of the foregoing target recognition methods.

[0139] Specifically, please refer to Figure 18 , Figure 18 which is a schematic block diagram of the target recognition device 100 provided by the embodiments of the present application.

[0140] As Figure 18 shown, the target recognition device 100 includes: an acquisition unit 110, a fusion unit 120, an identification unit 130, and a filtering unit 140.

[0141] The acquisition unit 110 is configured to acquire first recognition information of at least one lane line in the road and second recognition information of at least one obstacle around the vehicle traveling on the road;

[0142] The fusion unit 120 is configured to fuse the first recognition information and the second recognition information to obtain relative position information between the lane lines of the road and the obstacles around the vehicle;

[0143] The identification unit 130 is configured to determine target obstacles and non-target obstacles from the at least one obstacle according to the relative position information between the lane lines of the road and the obstacles around the vehicle and the motion states of the obstacles within a preset range around the vehicle, where the motion states of the obstacles include a stationary state and a moving state.

[0144] The target recognition device 100 provided by an embodiment of the present application is used to execute the above-mentioned acquisition of the first recognition information of at least one lane line on a road and the second recognition information of at least one obstacle around a vehicle traveling on the road; fuse the first recognition information and the second recognition information to obtain the relative position information between the lane line of the road and the obstacles around the vehicle; determine a target obstacle and a non-target obstacle from the at least one obstacle according to the relative position information between the lane line of the road and the obstacles around the vehicle and the motion states of the obstacles within a preset range around the vehicle, where the motion states of the obstacles include a stationary state and a moving state.

[0145] A filtering unit 140 is configured to filter at least one obstacle around the vehicle according to the non-target obstacle.

[0146] It should be noted that those skilled in the art can clearly understand that the specific implementation processes of the above-mentioned target recognition device 100 and each unit can refer to the corresponding descriptions in the foregoing method embodiments. For the sake of convenience and brevity of description, they will not be elaborated here.

[0147] The above-mentioned target recognition device can be implemented in the form of a computer program, and this computer program can run on an electronic device as shown in Figure 19 .

[0148] Please refer to Figure 19 , Figure 19 which is a schematic block diagram of an electronic device provided by an embodiment of the present application. The electronic device 200 can be a terminal. Among them, the terminal can be an electronic device with a communication function such as a smart phone, a tablet computer, a notebook computer, a desktop computer, a personal digital assistant, and a wearable device.

[0149] Refer to Figure 19 , the electronic device 200 includes a processor 202, a memory, and a network interface 205 connected through a system bus 201. Among them, the memory can include a non-volatile storage medium 203 and an internal memory 204.

[0150] The non-volatile storage medium 203 can store an operating system 2031 and a computer program 2032. The computer program 2032 includes program instructions, and when the program instructions are executed, the processor 202 can be made to execute a target recognition method.

[0151] The processor 202 is used to provide computing and control capabilities to support the operation of the entire electronic device 200.

[0152] The internal memory 204 provides an environment for the operation of the computer program 2032 in the non-volatile storage medium 203. When the computer program 2032 is executed by the processor 202, the processor 202 can be caused to execute a target recognition method.

[0153] The network interface 205 is used for network communication with other devices. Those skilled in the art can understand that Figure 19 the structure shown in is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the electronic device 200 to which the solution of this application is applied. The specific electronic device 200 may include more or fewer components than those shown in the figure, or combine some components, or have a different component arrangement.

[0154] Among them, the processor 202 is used to run the computer program 2032 stored in the memory to implement the following steps:

[0155] Obtain the first recognition information of at least one lane line in the road and the second recognition information of at least one obstacle around the vehicle traveling on the road; fuse the first recognition information and the second recognition information to obtain the relative position information between the lane line of the road and the obstacle around the vehicle; determine the target obstacle and the non-target obstacle from the at least one obstacle according to the relative position information between the lane line of the road and the obstacle around the vehicle and the motion states of the obstacles within a preset range around the vehicle, where the motion states of the obstacles include a stationary state and a moving state.

[0156] It should be understood that in the embodiment of this application, the processor 202 may be a central processing unit (CPU), and the processor 202 may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0157] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program includes program instructions, and the computer program can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0158] Therefore, the present application also provides a storage medium. The storage medium can be a computer-readable storage medium. The storage medium stores a computer program, where the computer program includes program instructions. When the program instructions are executed by a processor, the processor performs the following steps:

[0159] Obtain first recognition information of at least one lane line in a road and second recognition information of at least one obstacle around a vehicle traveling on the road; fuse the first recognition information and the second recognition information to obtain relative position information between the lane lines of the road and the obstacles around the vehicle; determine a target obstacle and a non-target obstacle from the at least one obstacle according to the relative position information between the lane lines of the road and the obstacles around the vehicle and the motion states of the obstacles within a preset range around the vehicle, where the motion states of the obstacles include a stationary state and a moving state.

[0160] The storage medium can be a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk, or an optical disc, etc., which are all computer-readable storage media that can store program codes.

[0161] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of the examples have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0162] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of each unit is only a logical function division, and there can 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.

[0163] The steps in the method of the embodiments of the present application can be adjusted in sequence, combined, and deleted according to actual needs. The units in the device of the embodiments of the present application can be combined, divided, and deleted according to actual needs. In addition, the functional units in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0164] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on such an understanding, the technical solution of the present application, in essence, 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. The computer software product is stored in a storage medium and includes several instructions for causing an electronic device (which can be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application.

[0165] As described above, the above are only specific implementation manners of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed in the present application can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A target recognition method, characterized in that, Including: Obtaining first identification information of at least one lane line in a road and second identification information of at least one obstacle around a vehicle traveling on the road; Fusing the first identification information with the second identification information to obtain relative position information between the lane line of the road and the obstacle around the vehicle; Determining a target obstacle and a non-target obstacle from the at least one obstacle according to the relative position information between the lane line of the road and the obstacle around the vehicle and the motion states of the obstacles within a preset range around the vehicle, wherein the motion states of the obstacles include a stationary state and a moving state; Filtering at least one obstacle around the vehicle according to the non-target obstacle to filter out the non-target obstacle and retain the target obstacle; Wherein, determining a target obstacle and a non-target obstacle from the at least one obstacle includes: determining a non-target obstacle and an obstacle to be corrected from the at least one obstacle, and correcting the obstacle to be corrected into a target obstacle or a non-target obstacle according to the size of the obstacle, and the obstacle to be corrected includes: an obstacle located on the lane line of the road and in a stationary state; 2. The object recognition method according to claim 1, wherein, The obtaining first identification information of at least one lane line in a road and second identification information of at least one obstacle around a vehicle traveling on the road includes: Obtaining first identification information of at least one lane line in a road according to a preset vision sensor; Obtaining second identification information of at least one obstacle around a vehicle traveling on the road according to a preset radar.

3. The object recognition method according to claim 1, wherein The obtaining first identification information of at least one lane line in a road includes: Obtaining initial lane line identification information including at least one lane line; Based on the initial lane line identification information, determining whether lane lines are identified on the left and right sides of the vehicle; If lane lines are identified on both the left and right sides of the vehicle, using the initial lane line identification information as the first identification information, otherwise, determining whether a lane line filling operation can be performed based on the initial lane line identification information; If the lane line filling operation can be performed, performing a lane line filling operation on the initial lane line identification information to obtain filled identification information, using the filled identification information as the first identification information, otherwise, using the initial lane line identification information as the first identification information; During the process of obtaining initial lane line identification information including at least one lane line, it further includes: correcting the lane line identification result of the current frame based on the lane line identification result of the previous frame; The correcting the lane line identification result of the current frame based on the lane line identification result of the previous frame includes: Obtaining third identification information of the lane line in the road at the current moment and fourth identification information of the lane line in the road at the previous moment; If the third recognition information does not include the recognition information of at least one lane line, then use the fourth recognition information as the initial lane line recognition information. Otherwise, match each lane line in the third recognition information among the multiple lane lines in the fourth recognition information, add at least one lane line in the fourth recognition information that does not match the lane lines in the third recognition information to the third recognition information, and use the third recognition information with the lane lines in the fourth recognition information added as the initial lane line recognition information.

4. The target recognition method according to claim 1, characterized in that The fusion of the first recognition information and the second recognition information to obtain the relative position information between the lane lines of the road and the obstacles around the vehicle includes: Based on a preset coordinate system, fuse the first recognition information and the second recognition information to obtain the first coordinate information of the lane lines of the road and the second coordinate information of the obstacles around the vehicle; Obtain the relative position information between the lane lines of the road and the obstacles around the vehicle according to the first coordinate information and the second coordinate information.

5. The target recognition method according to claim 1, characterized in that The motion state of the obstacle is obtained based on the second recognition information. The specific determination of the motion state of one obstacle is: Obtain the instantaneous speed of the obstacle from the second recognition information; Obtain the motion state of the obstacle according to the magnitude relationship between the instantaneous speed of the obstacle and a preset speed threshold. The motion state includes a stationary state and a moving state.

6. The object recognition method according to claim 1, wherein The motion state of the obstacle is obtained based on the second recognition information. The specific determination of the motion state of one obstacle is: Obtain the cumulative displacement of the obstacle within a preset time from the second recognition information; Obtain the motion state of the obstacle according to the magnitude relationship between the cumulative displacement of the obstacle and a preset displacement threshold. The motion state includes a stationary state and a moving state.

7. The target recognition method according to claim 1, wherein According to the relative position information between the lane lines of the road and the obstacles around the vehicle and the motion states of the obstacles within a preset range around the vehicle, determine the target obstacle and the non-target obstacle from the at least one obstacle, including: Determine the leftmost lane line and the rightmost lane line based on the first recognition information. Among the lane lines of the first recognition information, the lane line located on the left side of the vehicle and the farthest from the vehicle is the leftmost lane line, and the lane line located on the right side of the vehicle and the farthest from the vehicle is the rightmost lane line; Determine the non-target obstacle from the at least one obstacle. The non-target obstacle includes: an obstacle located on the left side of the leftmost lane line and with a distance greater than the first preset distance from the leftmost lane line, an obstacle located on the right side of the rightmost lane line and with a distance greater than the first preset distance from the rightmost lane line, an obstacle with a stationary motion state located on the left side of the leftmost lane line and with a distance less than the first preset distance and greater than the second preset distance from the leftmost lane line, and an obstacle with a stationary motion state located on the right side of the rightmost lane line and with a distance less than the first preset distance and greater than the second preset distance from the rightmost lane line; Determine a target obstacle from the at least one obstacle based on the non-target obstacle.

8. The target recognition method according to claim 2, wherein Determining a target obstacle and a non-target obstacle from the at least one obstacle according to the relative position information between the lane lines of the road and the obstacles around the vehicle and the motion states of the obstacles within a preset range around the vehicle includes: Determine the leftmost lane line and the rightmost lane line based on the first recognition information, where the lane line in the first recognition information that is on the left side of the vehicle and the farthest from the vehicle is the leftmost lane line, and the lane line in the first recognition information that is on the right side of the vehicle and the farthest from the vehicle is the rightmost lane line; Determine non-target obstacles from the at least one obstacle, where the non-target obstacles include: obstacles located on the left side of the leftmost lane line and with a distance greater than a first preset distance from the leftmost lane line, obstacles located on the right side of the rightmost lane line and with a distance greater than a first preset distance from the rightmost lane line, obstacles with a motion state of stationary state that are located on the left side of the leftmost lane line and with a distance less than the first preset distance and greater than a second preset distance, and obstacles with a motion state of stationary state that are located on the right side of the rightmost lane line and with a distance less than the first preset distance and greater than a second preset distance; Determine an obstacle to be corrected from the at least one obstacle, where the obstacle to be corrected includes: an obstacle located on the lane line of the road and with a motion state of stationary state; Correct the obstacle to be corrected into a target obstacle or a non-target obstacle; Determine a target obstacle from the at least one obstacle based on the non-target obstacle, the corrected non-target obstacle, and the corrected target obstacle.

9. The object recognition method according to claim 8, characterized in that The correcting the obstacle to be corrected into a target obstacle or a non-target obstacle includes: Obtain the radar cross section of the obstacle to be corrected from the second recognition information; Correct the obstacle to be corrected into a target obstacle or a non-target obstacle according to the size relationship between the radar cross section of the obstacle to be corrected and a preset cross section threshold.

10. A target recognition device, characterized in that, Includes: An acquisition unit for acquiring first recognition information of at least one lane line in the road and second recognition information of at least one obstacle around a vehicle traveling on the road; A fusion unit for fusing the first recognition information and the second recognition information to obtain relative position information between the lane lines of the road and the obstacles around the vehicle; An identification unit for determining a target obstacle and a non-target obstacle from the at least one obstacle according to the relative position information between the lane lines of the road and the obstacles around the vehicle and the motion states of the obstacles within a preset range around the vehicle, where the motion states of the obstacles include a stationary state and a moving state; The target recognition device is further configured to filter at least one obstacle around the vehicle according to the non-target obstacle to filter out the non-target obstacle and retain the target obstacle; Among them, determining the target obstacle and non-target obstacles from the at least one obstacle includes: determining non-target obstacles and obstacles to be corrected from the at least one obstacle, and correcting the obstacles to be corrected into target obstacles or non-target obstacles according to the size of the obstacles. The obstacles to be corrected include: obstacles located on the lane line of the road and in a stationary state of motion.

11. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the target recognition method according to any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which when executed by the processor causes the processor to execute the target recognition method according to any one of claims 1 to 9.

Images