Efficient Foreground Layer Extraction in Video

Extracting foreground moving objects from video sequences is an important task and also a hot topic in computer vision and image processing. Segmentation results can be used in many object-based video applications such as object-based video coding, content-based video retrieval, intelligent video surveillance, video-based human-computer interaction, etc. In this paper, we propose a framework for real-time segmentation of foreground moving objects from monocular video sequences with static background. Our algorithm can extract foreground layers with cast shadow removal accurately and efficiently. To reduce the computation cost, we use Gaussian Mixture Models to model the scene and obtain initial foreground regions. Then we combine the initial foreground mask with shadow detection to generate a quadrant-map for each region. Based on these quadrant-maps, Markov Random Field model is built on each region and the graph cut algorithm is used to get the optimal binary segmentation. To ensure good temporal consistency, we reuse previous segmentation results to build the current foreground model. Experimental results on various videos demonstrate the efficiency of our proposed method.