In this paper an automatic detection and segmentation algorithm of video multiple moving targets is proposed for the problem of computer vision in intelligent monitoring system. The algorithm improved the adaptive clustering by defining the pixel spatial connectivity rate. We design the perpendicular split method, initial cluster adaptive splitting and merging self-organizing the iterative clustering segmentation algorithm. It improved the active contour model to complete the edge detection. Experimental results show that multiple moving targets segmentation results are consistent with the human visual judgment, take use of space connectivity information improves the accuracy of clustering segmentation, take use of sparse matrix block operation for active contour model that improves multiple moving targets edge detection result, comparison and analysis the experimental results show that the proposed algorithm is feasible, rapid and effective.
Download Full PDF Version (Non-Commercial Use)