A Visual Tracking Algorithm Based on Histogram of Gradient Feature

Particle Filter is a Monte Carlo method that designed to approximate nonlinear problem. It usually used to track the target of the dynamic system only based on the color feature. The single characteristic is vulnerable to the impact on background noise, light, and some other factors. These would lead to misplace or lose the target. In order to improve the efficiency and robustness of the object tracking in complex scenes, an approach by using Histogram of Oriented Gradient (HOG) is presented under the Particle Filter framework. This method characterizes the tracked objects using HOG features. While the gradient descriptor is not sensitive to light, its robustness and applicability of the object tracking method can be increased. The global approach has been tested on different test videos and real-world sequences. Experimental results show the robustness and efficiency of the proposed method in several difficult scenarios, especially compared with the traditional particle filter algorithm.