Improved Multi-object detection and tracking method based on mean shift algorithm

Abstract

Based on the Improved Mean Shift algorithm, connected domain search algorithm and Kalman filtering algorithms, a multiple targets tracking algorithm is proposed. It is good at solving the target lost, overlapp and the object-clustering problem in the process of tracking. It could adapt its search window size and direction. By adaptive extending the window size, searching the information of the target orthocenter and size solves the target lost problem caused by the targets moving too fast. Moreover, Kalman filter is used for estimating the position of each moving object to overcome the moving objects interleaving or ovrelapping, and noise disturbance. The experimental results show that the proposed algorithm can improve the performance of multi-object tracking effectively, and realize keeping on multi-object tracking continuously. © 2011 International Information Institute.