Vehicle tracking threshold updating method and device, electronic equipment and storage medium

Abstract

The invention provides a vehicle tracking threshold value updating method and device, electronic equipment and a storage medium, and relates to the technical field of image processing, and the method comprises the steps: obtaining a to-be-processed image of a vehicle on a monitored road, determining the position information of a tracking target in the image according to a pre-trained detection model, and determining the tracking threshold value of the vehicle according to the position information. And updating the intersection-to-union ratio threshold value, so as to track the tracking target based on the updated intersection-to-union ratio threshold value. According to the updating mode combining position information monitoring and the intersection-to-parallel ratio threshold value, the problem of ID hopping of nonlinear geometric scale data association tracking can be solved, and compared with an existing Kalman filtering method, the error rate of data association is reduced, and the actual application effect of target tracking is improved.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a vehicle tracking threshold updating method, device, electronic equipment and storage medium. Background technique [0002] The visual tracking system can automatically discover and track the selected target from the original video collected by the camera. This system is one of the basic technologies in the field of computer vision and is widely used in vehicle monitoring, unmanned driving, robots, security monitoring, etc. application. [0003] In the intelligent road traffic system, the tracking and positioning of road vehicle driving information is the key technology of intelligent traffic management research. At this stage, vehicle tracking technology mostly relies on vehicle GPS tracking system, 3D lidar, camera video target detection and target tracking based on artificial intelligence, which greatly promotes the intelligent development of off-site highwa...

AI Technical Summary

Problems solved by technology

[0004] However, when it is necessary to track a target vehicle from the traffic video stream, the traditional manual monitoring method is not only time-consuming and laborious, but also easily misses key video information, which is inefficient.

And in the single camera scene, there are shortcomings of too small monitoring range and insufficient information

Although the multi-camera method can make up for the shortcomings of a single camera, the existing "cross-mirror" tracking technology is not mature enough, the tracking ID jumps frequently, and the stability is poor. Most practical applications still rely on manual clicks to switch between videos. Its applicability is not high

The multi-eye panoramic camera stitching can also provide a wide range of field of view, but the stitching imaging process introduces short-range nonlinear spherical distortion, resulting in "non-linear scale changes" of vehicle targets, which has a significant impact on the continuous tracking of variable-range targets

Images