A background extraction and shadow removal algorithm based on clustering for ViBe
Yin-Shi Qin, Shuifa Sun, Xian-Bing Ma, Song Hu, Bangjun Lei · 2014
Shadow removal has always been one of the hot research topics in the fields of computer vision. Recently the ViBe (Visual background extraction) foreground extraction algorithm which is based on probability and statistics gets more and more attention due to its high speed and simplicity. However, its biggest drawback is the poor performance for videos with moving shadows. Because the process of the original ViBe algorithm is carried out in the gray space, it is against the extraction of background as existing shadow removal algorithms typically operate with a colored background image. In this paper, a background extraction and shadow removal algorithm based on clustering for ViBe is proposed. Firstly, the color background is extracted by clustering the values of each pixel in R, G, B components space. Then the background information is used to remove shadow hidden in the foreground. For indoor and outdoor videos with moving cast shadows, ROC (Receiver Operating Characteristic) curve is used to validate the proposed approach. Experimental results show that a good performance has been gained in shadow removal.