Robust object tracking via improved mean-shift model

Abstract

In this paper we propose a robust object tracking algorithm using a improved Mean-Shift model. As the traditional Mean-Shift algorithm for object tracking uses a single histogram. Because the traditional Mean-Shift lacks spatial distribution information, so it is difficult to track non-rigid object especially. With a focus on this problem, an improved Mean-Shift algorithm based on the shape feature and color of the target is presented. The results show that the algorithm can track the moving vehicles in real time, and it has a preferable adaptability and robustness to the irregular motion and deformation of the target.