Abstract:Moving objects detection is one of fundamental tasks of video surveillance.Specific to the poor identification capability and low accuracy to shadow in gray information,this paper proposes a novel moving objects detection method based on the combination of Low-Rank Matrix decomposition and HSV color information.Firstly,we convert the images from RGB space to HSV space,construct observation matrix for H,S,V channels,respectively,and optimize the observation matrix through Low-Rank Matrix decomposition to obtain H,S,V channel's foreground component and background component;combing H,S,V channels foreground component in order to get roughly moving object district.Secondly,the moving shadow should be detected and eliminated from the foreground image,after combining H,S,V channels component to get the row processing foreground objects and the column processing foreground objects,the row processing foreground image and the column processing foreground image are combined to obtain the moving objects image.Finally,by morphological processing and connectivity detection to eliminate the noise,the accurate foreground moving objects can be obtained.The experimental results demonstrate that the proposed method is much better than others in increasing accuracy of moving objects detection.