A Robust Mean Shift Tracking Algorithm Combined with Gray Spatial Texture

Abstract

The classical Mean Shift algorithm has some limitations in target tracking of gray image sequences for it only adopts the simple gray color information. We present an improved Mean Shift tracking algorithm in this paper. Our approach employs the information of gray spatial texture which is used to characterize the object in matching models calculation. Thus, the proposed algorithm can recognize the real target from a complicated background. The experiments show that improved Mean Shift tracking algorithm has higher efficiency and reliability than previous method.