Research on Video Object Tracking Based on Improved Camshift Algorithm
Zhenxing Fu, Peijiang Chen, Yaochang Xi · 2022
Target tracking is one of the important research directions of computer vision, widely used in intelligent surveillance, pedestrian detection, military war, and other fields. In this paper, the Camshift algorithm is improved based on multi-feature fusion. Aiming at the problem that the color of the tracking target is similar to that of the background, the Local Binary Pattern feature is partly improved and adaptively fused with the template generated by the joint chromaticity and saturation components. As it is easy to lose the target in the case of occlusion, the improved algorithm incorporates the Kalman filter. According to the statistics of the experimental results, the improved algorithm achieves a better tracking effect.