Approach of moving objects detection in active video surveillance

Image multi-scale edge detection based on anti-symmetrical biorthogonal wavelet is given detailedly in theory, Especially an algorithm of wavelet decomposition in which multi-scale edge can be detected is put forward. Based on above using different kinds of wavelet decomposition data such as approximate coefficients and multi-scale edge image, a moving object detection approach in active video surveillance system is proposed. Firstly Block Matching is used to calculate motion vector in low frequency image of wavelet decomposition. Then Motion Compensation and Frame Difference between two adjacent frames applied to extract moving object information. And using multi-scale edge image of wavelet Region Growth is combined with Hough Transformation to define region of potential object. Finally object is farther segmented in object potential region. Experimental results indicate that the method presented in the paper is effective.