Visual multi-target tracking method based on multiple single trackers

Abstract

The invention discloses a visual multi-target tracking method based on multiple single trackers. The visual multi-target tracking method comprises the steps that targets are detected based on a classifier and are tracked through the multiple visual single-target trackers, the targets and the trackers are corrected through different strategies during tracing, and thus the multiple targets are tracked continuously; specifically, firstly, the target classifier is utilized to detect images so as to obtain the targets, then one visual single-target tracker is assigned to each target, and the multiple visual single-target trackers are utilized to jointly complete the task of multi-target tracking; and in order to deal with the cumulative error of the trackers, targets escaping from the view during tracking and new targets, the update cycle for the multiple single-target trackers and the multi-tracker consistency judgment cycle are introduced, thus adaptive management such as update and consistency judgment is conducted cyclically on the multiple single-target trackers, and the multiple targets are effectively tracked. According to the visual multi-target tracking method, the tracking efficiency is greatly improved, and the requirement for multi-target real-time tracking is basically met.

Description

technical field [0001] The invention belongs to the field of visual multi-target tracking, in particular to a visual multi-target tracking method based on multiple single trackers. Background technique [0002] Visual multi-target tracking technology is one of the key technologies of computer vision. It is widely used in video surveillance, disaster scene search and rescue, military target strike and emerging automatic driving and other fields. [0003] At present, the vast majority of research at home and abroad is on visual single-target tracking, and there are relatively few researches on visual multi-target tracking. The existing visual multi-target tracking algorithms are basically inseparable from the data association link. First, the target classifier is used to detect and obtain the target, and then the target is tracked and associated based on the data association algorithm to achieve multi-target tracking. The huge computing load cannot meet the requirements of mu...

AI Technical Summary

Problems solved by technology

The existing visual multi-target tracking algorithms are basically inseparable from the data association link. First, the target classifier is used to detect and obtain the target, and then the target is tracked and associated based on the data association algorithm to achieve multi-target tracking. Huge computing load, unable to meet the requirements of multi-target real-time tracking

There are several multi-target tracking methods that use the confidence of tracking segments and discriminative appearance learning. This method performs local data association between tracking segments and detection results or global data association between tracking segments based on the confidence of tracking segments. Appearance module learning to distinguish different objects in the tracking process; this method is also inseparable from the data association process, and its tracking frame rate is usually lower than 5 frames per second; In this situation, the appearance of the target changes significantly, and it is difficult to obtain a robust target appearance model

There is also a video multi-target tracking method based on joint reasoning. In this method, the video frame image is first rasterized, and then the online detector and the KLT (Kanade-Lucas-Tomasi Tracking) tracking algorithm are used to calibrate the candidate position of the target. The obtained Candidate position results are quantified and scored, and finally the joint function is used to describe the target tracking situation, and the optimal solution based on the joint function is used as the position of the target in this frame, that is, the tracking of the target is realized; in this method, the quantitative score of the candidate target and The optimal solution process of the joint function is essentially a data association process. The KLT tracking algorithm and the joint function solution have high time complexity and it is difficult to achieve real-time performance (the tracking frame rate is usually lower than 5 frames per second).

Images