A multi-target tracking method based on feature point trajectory

Abstract

The invention discloses a multi-target tracking method based on feature point trajectory, and relates to the technical field of multi-target tracking. The invention includes a generative model, which describes the apparent features of the target by establishing a model; a discriminant model, which uses a classifier to discriminate whether the image block belongs to the target or the background; and multi-target tracking, which adopts an offline tracking method in a small time segment, that is, in a small time segment. Offline tracking is performed within N frames; multi-target tracking also includes acquiring feature points of N frames of images in advance, feature point trajectories, and measuring tracking accuracy. The invention obtains the feature points of the image in advance through multi-target tracking, uses the future information to guide the tracker, searches for the features of the target in a larger time span, and helps to solve the occlusion problem of target tracking; and uses the feature point trajectory to track the target, Solve the problem of feature points disappearing. The target is occluded in the current frame. When the occlusion is eliminated, the feature points will be matched with the trajectory again, which will not affect the tracking effect.

Description

technical field [0001] The invention belongs to the technical field of multi-target tracking, in particular to a multi-target tracking method based on feature point trajectory. Background technique [0002] In complex scenes, an effective and robust multi-target tracking algorithm is the basis for many computer vision applications, such as video surveillance, autonomous driving, and human-computer interaction. The task of multi-target tracking is to demarcate the target object with a bounding rectangle and determine its specific position in the next video frame. [0003] In the case of occlusion, the existing tracking device is easy to introduce the occlusion area into the model and update it incorrectly, which will eventually lead to the error and drift of the tracker, which will affect the tracking effect. SUMMARY OF THE INVENTION [0004] The purpose of the present invention is to provide a multi-target tracking method based on the feature point trajectory, through mul...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a multi-target tracking method based on feature point trajectory, through multi-target tracking and early acquisition of feature points of the image, it solves the problem that the existing tracking device easily introduces the occlusion area into the model when occlusion occurs In the middle and wrong update, the error and drift of the tracker will be caused, which will affect the tracking effect.

Images