The motion target detection algorithm for visual background extraction (ViBe) is widely used in dynamic target detection in video sequences, but because of the ghosting and shadow problems in the traditional ViBe algorithm, it will have some influence on the accuracy of target detection. In view of the problem that the traditional ViBe algorithm is difficult to eliminate in detecting the ghost phenomenon in the detection target, a method of combining the improved three-frame difference method, the maximum inter-class variance method (Otsu) and the ViBe algorithm is proposed to detect and eliminate ghost problems, and a method combining HSV color space and texture characteristics is proposed to detect and eliminate the shadow. The simulation experiments on CDW-2014, the public video database, show that the algorithm can not only effectively remove ghost and shadow problems, but also extract the moving target accurately.