Human detection in video surveillance using MBCCA: Macro Block Connected Component Algorithm

The detection of incredulous behavior of human in public video surveillance has captivated an increasing level of concentration in computer vision. Video surveillance is the process of scrutinizing the video sequences. In this system there is no human intercession. To monitor a video for a long endurance by a human operator is highly speculative and futile. This paper deals with automatic detection of human in public video surveillance. In this paper, the proposed MBCCA (Macro Block Connected Component Algorithm) is used to detect and segment the human in a video. MB classification classifies the image frame into macro blocks to identify the moving regions by computing motion vector for each block. The detected moving region is given as input to SVM classifier to classify the human from non-human objects. Then the block containing moving human is segmented using connected component algorithm. The proposed MBCCA is more efficient than the existing MB classification and connected component algorithm where the entire image is given as input for segmentation using connected component algorithm.