Algorithm of moving object detection based on multifeature fusion
Jianrong Cao, Xuemei Sun, Shusheng Zhao, Yameng Wang, Shulan Gong · 2017
In view of the foreground and background similarity of building in some monitoring video and the caliginous environment light, the foreground segmentation is not complete. This paper proposes a background modeling algorithm combining color feature and XCSLBP texture feature. When the color of foreground and background is similar, the foreground is detected by the difference of the texture feature. When the texture of foreground and background information is similar, the foreground detection is performed by the difference of the color feature. In this method, the Gaussian mixture model is constructed by the color feature, and the XCSLBP texture operator is used to construct the texture background model. The experimental results show that our method has the higher foreground detection accuracy compared with the single feature detection methods.