Improved Online Boosting and Kalman filter improvement-based TLD tracking method

Abstract

The invention discloses an improved Online Boosting and Kalman filter improvement-based TLD tracking method, and belongs to the technical field of machine vision, artificial intelligence, man-machineinteraction and target tracking. The method comprises the following steps of: (1) initialization: initializing an improved Online Boosting classifier and a P-N learning device by utilizing an initialsample set formed through selecting a target and carrying out affine transformation; (2) image tracking: selecting a feature point, tracking the feature point for twice by using an L-K optical flow method, and comparing an error between the twice tracking with a threshold value so as to obtain a tracking result; (3) image detection: obtaining a detector result through a Kalman filter, a variance classifier, the Online Boosting classifier and a KNN classifier; (4) tracking result and detection result integration: assessing confidence coefficients of the tracker result and the detector result soas to determine which module result is finally adopted; and (5) online learning: correcting the tracker result and the detector result by using the P-N learning device, and enriching the sample set.The method is capable of effectively overcoming the shielding problems, improving the speed of original methods and effectively the precision and robustness of detectors.

Description

technical field [0001] The invention relates to a TLD (Tracking-Learning-Detection tracking-learning-detection) tracking method based on improved Online Boosting (online cascade classifier) ​​and Kalman filter improvement, belonging to machine vision, artificial intelligence, human-computer interaction and Object tracking technology field. Background technique [0002] Video image tracking has always been the focus of attention in the field of computer and image. The early video tracking mainly used the target tracking technology based on feature matching, which mainly used the light and shade, edge, color, texture and temporal and spatial differences of the moving target in the image sequence to detect the moving object. Among them, the literature (Meanshift proposed by Comaniciu D, Meer P: a robust method for feature space analysis, published in the direction of IEEE pattern recognition and machine intelligence) and the literature (Allen J G, Xu R Y D, Jin J S proposed th...

AI Technical Summary

Problems solved by technology

However, when the target is occluded by a large area, the accuracy of the target tracking is severely reduced by this method, and when the initial tracking is performed, there are fewer samples, and the robustness of the detector is poor.

Images