Color Comparison in Codebook Model for Moving Objects Detection

For surveillance video, color information in scenes is important for detecting moving objects. This paper focuses on the selection of color space and the comparison of color distortion in codebook model which represents each background pixel with some codewords based on its color distortion and brightness bounds. Several computing methods are experimented for contrast. The results show that HSV color space achieves better effectiveness in detecting moving objects for a certain surveillance video.