CN-121777945-B - Method for determining course angle of vehicle and computer equipment

CN121777945BCN 121777945 BCN121777945 BCN 121777945BCN-121777945-B

Abstract

The application provides a method for determining a vehicle course angle and computer equipment, the method comprises the steps of determining characteristic information of a plurality of target detection areas of a target vehicle based on an original traffic environment image corresponding to a traffic environment where the target vehicle is located, determining course angle information and confidence coefficient information corresponding to each target detection area according to the characteristic information of each target detection area, determining weights corresponding to each course angle information according to the confidence coefficient information corresponding to each target detection area, and obtaining target course angle information of the target vehicle according to the course angle information corresponding to each target detection area and the weights corresponding to each course angle information. Therefore, for a single-frame original traffic environment image, the target course angle information of the vehicle can be obtained according to the course angle information corresponding to a plurality of target detection areas of the vehicle and the weight corresponding to each course angle information determined according to the corresponding confidence information, the real-time requirement can be met, and the accuracy of the course angle of the vehicle is guaranteed or improved.

Inventors

FANG ZHIMING
LIU XIN
WU GUOWEI

Assignees

纽劢科技(上海)有限公司

Dates

Publication Date: 20260512
Application Date: 20260309

Claims (9)

1. A method of determining a heading angle of a vehicle, the method comprising: Acquiring an original traffic environment image corresponding to a traffic environment in which a target vehicle is located, and detecting and normalizing the original traffic environment image to determine a vehicle image area of the target vehicle from the original traffic environment image; Determining characteristic information of a plurality of target detection areas on the target vehicle according to the vehicle image area, wherein the plurality of target detection areas comprise areas respectively corresponding to a vehicle body frame, double lamps, single tail lamps and license plates, and the characteristic information corresponding to the plurality of target detection areas comprises vehicle body frame characteristic information, double lamp characteristic information, single tail lamp characteristic information and license plate characteristic information; Determining the course angle information corresponding to each target detection area and the confidence coefficient information corresponding to each course angle information according to the characteristic information of each target detection area, wherein the course angle information corresponding to each target detection area and the confidence coefficient information corresponding to each course angle information are determined according to the characteristic information of each target detection area and comprise the steps of determining the first course angle information corresponding to the car body frame and the car body frame confidence coefficient information corresponding to the first course angle information according to the car body frame characteristic information, determining the second course angle information corresponding to the double lamps and the double lamp confidence coefficient information corresponding to the second course angle information according to the double lamp characteristic information, determining the third course angle information corresponding to the single lamp and the single lamp confidence coefficient information corresponding to the third course angle information according to the single lamp characteristic information, and determining the fourth course angle information corresponding to the license plate and the license plate confidence coefficient information corresponding to the fourth course angle information according to the license plate characteristic information; determining the weight corresponding to each course angle information according to the confidence coefficient information corresponding to each course angle information; and determining target course angle information of the target vehicle according to the course angle information corresponding to each target detection area and the weight corresponding to each course angle information.
2. The method for determining a heading angle of a vehicle according to claim 1, wherein the vehicle body frame feature information includes a set of vehicle body contour pixel points, and determining, based on the vehicle body frame feature information, first heading angle information corresponding to the vehicle body frame and vehicle body frame confidence information corresponding to the first heading angle information includes: Determining centroid coordinate information of the target vehicle according to the vehicle body contour pixel point set; Determining a covariance matrix according to the vehicle body contour pixel point set and the centroid coordinate information; Determining a minimum eigenvalue, a maximum eigenvalue and an eigenvector corresponding to the maximum eigenvalue of the covariance matrix; determining the first course angle information according to the feature vector, and And determining the confidence information of the vehicle body frame according to the ratio of the maximum characteristic value to the minimum characteristic value.
3. The method for determining a heading angle of a vehicle according to claim 1, characterized in that, The double-lamp characteristic information comprises lamp color information, lamp brightness information and lamp symmetry degree information, and the method further comprises the following steps: Determining lamp category information of the double lamps according to the lamp color information, the lamp brightness information and the lamp symmetry degree information, wherein the lamp category information of the double lamps comprises the lamp serving as a head lamp or the lamp serving as a tail lamp; The double-lamp feature information further comprises first lamp position information, second lamp position information, lamp detection confidence, lamp symmetry confidence and lamp color confidence, and the second course angle information corresponding to the double lamps and the double-lamp confidence information corresponding to the second course angle information are determined according to the double-lamp feature information, and the double-lamp feature information comprises: determining a car light connection vector according to the first car light position information and the second car light position information; Determining the second course angle information corresponding to the direction perpendicular to the car light connection vector according to the car light connection vector, wherein if the double-car light is a car light, the second course angle information is the second course angle information corresponding to the car light, and if the double-car light is a car tail light, the second course angle information is the second course angle information corresponding to the car tail light, and And determining double-lamp confidence information corresponding to the second course angle information according to the lamp detection confidence, the lamp symmetry confidence and the lamp color confidence.
4. The method for determining a heading angle of a vehicle according to claim 1, wherein the single-taillight feature information includes a taillight pixel area, and determining third heading angle information corresponding to the single-taillight and single-taillight confidence information corresponding to the third heading angle information according to the single-taillight feature information includes: based on an image detection method, obtaining a long-side direction vector of the tail lamp according to the tail lamp pixel area; Determining the third course angle information corresponding to the direction perpendicular to the long-side direction vector of the tail lamp according to the long-side direction vector of the tail lamp, and And determining the single tail lamp confidence information corresponding to the third course angle information according to the tail lamp pixel area.
5. The method for determining a heading angle of a vehicle according to claim 1, wherein the license plate feature information includes position information of four corner points of the license plate, license plate detection confidence information and shielding flag information, and determining fourth heading angle information corresponding to the license plate and license plate confidence information corresponding to the fourth heading angle information according to the license plate feature information includes: determining first length information and second length information of the license plate according to the position information of the four corner points; Determining the perspective proportion of the license plate according to the first length information and the second length information; determining the fourth course angle information according to the perspective proportion, and And determining license plate confidence information corresponding to the fourth course angle information according to the license plate detection confidence information and the shielding mark information.
6. The method of determining a heading angle of a vehicle according to claim 1, characterized in that determining feature information of a plurality of target detection areas on the target vehicle from the vehicle image area includes: Performing edge detection processing on the vehicle image area to obtain the vehicle body frame characteristic information on the target vehicle, or performing semantic segmentation processing on the vehicle image area based on a semantic segmentation model to obtain the vehicle body frame characteristic information of the vehicle body frame; performing double-lamp key point detection processing on the vehicle image area based on a key point detection model to obtain the double-lamp characteristic information corresponding to the double lamps; based on an image detection model or the semantic segmentation model, performing single-taillight detection processing on the vehicle image area to obtain the single-taillight characteristic information corresponding to the single-taillight; And carrying out license plate detection processing on the vehicle image area based on a license plate detection model to obtain license plate characteristic information corresponding to the license plate.
7. The method for determining a heading angle of a vehicle according to claim 1, characterized in that, The method further comprises the steps of: determining a scene type corresponding to the traffic environment where the target vehicle is located according to the original traffic environment image; performing geometric verification on the course angle information corresponding to each target detection area, and determining a geometric consistency verification result of the course angle information corresponding to each target detection area; determining the weight corresponding to each course angle information according to the confidence information corresponding to each course angle information, including: And determining the weight corresponding to each course angle information according to the confidence information corresponding to each course angle information, the scene type and the geometric consistency check result.
8. The method for determining a heading angle of a vehicle according to any one of claims 1 to 7, characterized in that detecting and normalizing the original traffic environment image to determine a vehicle image area of the target vehicle from the original traffic environment image includes: performing image detection processing on the original traffic environment image based on an image detection model to determine a detection frame of the target vehicle from the original traffic environment image; And performing size normalization processing on the detection frame of the target vehicle to determine the vehicle image area of the target vehicle from the original traffic environment image.
9. A computer device for performing the method of determining a heading angle of a vehicle as claimed in any one of claims 1-8.

Description

Method for determining course angle of vehicle and computer equipment Technical Field The present application relates to the field of vehicle control technologies, and in particular, to a method for determining a heading angle of a vehicle and a computer device. Background Along with the development of vehicle auxiliary driving technology, the requirements for vehicle running track prediction, collision prediction and planning control are higher and higher. Currently, a vehicle is generally required to determine a heading angle of a target vehicle in front of or around the host vehicle, so as to predict a running track of the target vehicle, pre-judge collision between the host vehicle and the target vehicle, formulate a planning control strategy of the host vehicle, and the like based on the determined heading angle of the target vehicle. Therefore, the accuracy of the heading angle of the target vehicle influences the accuracy of the prediction of the running track of the vehicle, the collision pre-judgment and the planning control. In the prior art, the heading angle of a target vehicle is estimated based on the movement direction of the target vehicle in a plurality of consecutive frames. On the other hand, for stationary vehicles or vehicles running slowly, because the change of the moving direction of continuous multiframes is tiny or even unchanged, the effective course angle cannot be estimated, and the accuracy of the obtained course angle is affected. Therefore, the existing method for determining the course angle of the target vehicle has the problems that the single-frame real-time requirement cannot be met and the accuracy of the course angle is affected. Disclosure of Invention The application provides a method for determining a vehicle course angle and computer equipment, which are used for determining characteristic information of a plurality of detection areas of a target vehicle based on an original traffic environment image corresponding to a traffic environment where the target vehicle is located, determining course angle information and confidence coefficient information corresponding to each detection area according to the characteristic information of each detection area, and adaptively adjusting weights corresponding to each course angle information according to the confidence coefficient information corresponding to each detection area so as to obtain the target course angle information of the target vehicle according to the course angle information corresponding to each detection area and the weights corresponding to each course angle information. Therefore, for a single-frame original traffic environment image, the course angle information corresponding to a plurality of detection areas of the vehicle can be obtained according to the characteristic information of the plurality of detection areas of the vehicle, so that the target course angle of the vehicle is obtained, and the real-time requirement can be met. In addition, because the course angle information corresponding to different areas of the vehicle is considered, and the weight corresponding to each course angle information can be adaptively adjusted based on the confidence degree information corresponding to each detection area, the finally obtained target course angle information is more accurate, and the accuracy of the course angle of the vehicle is improved. In a first aspect, an embodiment of the application provides a method for determining a heading angle of a vehicle, which includes obtaining an original traffic environment image corresponding to a traffic environment where a target vehicle is located, detecting and normalizing the original traffic environment image to determine a vehicle image area of the target vehicle from the original traffic environment image, determining feature information of a plurality of target detection areas on the target vehicle according to the vehicle image area, determining heading angle information corresponding to each target detection area and confidence information corresponding to each heading angle information according to the feature information of each target detection area, determining weights corresponding to each heading angle information according to the confidence information corresponding to each heading angle information, and determining the target heading angle information of the target vehicle according to the heading angle information corresponding to each target detection area and the weights corresponding to each heading angle information. By adopting the technical scheme, the vehicle image area of the target vehicle is obtained by detecting and normalizing the acquired original traffic environment image corresponding to the traffic environment of the target vehicle, the characteristic information of a plurality of target detection areas on the target vehicle is determined according to the vehicle image area of the target vehicle, the course angle information a