Target detection tracking method based on TLD algorithm

Abstract

The invention discloses a target detection tracking method based on a TLD. The method comprises the following steps of (1) at an initial frame of a video to be tracked, a tracking window assigned by a user forms a positive and negative sample pair detection module to carry out initialization training; (2) during a tracking process, a detection module and a tracking module work independently; the detection module calculates a LBP statistics characteristic vector of an image block accepted by a variance classifier and a merging classifier, carries out NCC nearest neighbor matching with a sample set and outputs a target position; and the tracking module uses a median optical flow method to predict a target position of a current frame through previous frame tracking; (3) an integration module integrates the detection module and the tracking module so as to carry out tracking result output, and for the currently-updated target position, a new positive and negative sample is generated so as to update the detection module; and (4) the step (2) and the step (3) are circulated till that tracking is ended. A comparison experiment performed on a public data set shows that by using the method, robustness and precision of the tracking can be increased to some extent.

Description

technical field [0001] The invention relates to a target detection and tracking method based on the TLD algorithm, which is characterized in that the image of the scanning window in the detection module of the TLD algorithm is extracted and counted for LBP features, and is applied to computer vision, target detection, target tracking, etc., and belongs to the computer The field of object tracking in vision. Background technique [0002] The TLD tracking algorithm is a single target long-term tracking algorithm proposed by Zdenka Kalal. The algorithm consists of three modules: tracking module, detection module and learning module. It is difficult for a simple tracking algorithm to correct tracking drift errors and will continue to accumulate tracking errors, and once the target disappears from the field of view, tracking will inevitably fail. The simple detection algorithm requires a large number of samples for offline supervised training, and cannot be applied to the track...

AI Technical Summary

Problems solved by technology

[0008] The technical problem to be solved in the present invention is: aiming at the problem of insufficient robustness and low precision caused by the nearest neighbor classifier directly performing NCC matching on pixels in the detection module of the TLD algorithm, the detection module of the algorithm is improved, and the Use the statistical feature vector of LBP to reconstruct the detection module

Images