Target Tracking Based on Camshift Algorithm and Multi-feature Fusion

Abstract

Abstract This paper proposes a new algorithm on the basics of Camshift algorithm and multi-feature fusion. First, the target area is weighted by the Gaussian function. Then LBP is used to extract local texture features, next fuse the texture features with the color features to obtain the color and texture histogram. Finally, for the purpose of tracking the target accurately and preventing occlusion, the Kalman filter is used to forecast the position of the moving target. From the comparison of the experimental results, the conclusion can be drawn that the new algorithm can effectively overcome the interference of objects and track the target accurately in a more complex environment.