Object Tracking Algorithm based on Multi-channel Extraction of AHLBP Texture Features

Abstract

Aiming at the problem of tracking bias and deteriorating robustness due to background interference and sudden change of object movement state in video tracking, this study proposed a new object tracking algorithm based on multichannel extraction of AHLBP texture features. This algorithm took into account the local visual information of the image, proposed an AHLBP texture extraction method with adaptive threshold. The method overcame the limitations caused by a single fixed threshold in texture extraction. In order to mine image texture information in depth, textures were extracted from the three color channels in the HSV color space. Then the paper fully used the texture to establish an accurate and complete object model as a description of the object, and achieved object tracking. Experimental results show that the algorithm can effectively improve the accuracy and stability of tracking and make the tracking algorithm has a better robustness.