Automatic human body segmentation using level-set based active contours followed by optical flow in video surveillance

Human body segmentation is a critical module in video-based activity recognition (AR) because it defines the image area necessary and sufficient for the follow-up modules like feature extraction. Existing methods often involve modeling of the human body and/or the background, which normally requires extensive amount of training data and cannot efficiently handle changes over time. Recently, active contours have been emerging as an effective segmentation technique in still images. In this paper, an active contour model is adapted that is robust to illumination and clothing changes, typical issues in practical AR systems. To make the model work smoothly with video data, the optical flow is used, which is estimated in two consecutive frames, to position the initial contour in the current frame. The proposed approach is unsupervised, i.e., no training data or prior human model is needed. The proposed model gives prominent results of segmentation.