A lightweight distributed video coding scheme is presented for video surveillance applications. Background frames and foreground maps are generated using a video analytics module on the decoder side. Only sub-frames containing foreground blocks are encoded using a transform-domain Wyner-Ziv codec. A decoded frame is synthesised with the reconstructed sub-frames and the background frame. Through a simulation, the proposed scheme is shown to achieve better than state-of-the-art coding efficiency while reducing coding complexity greatly both at the encoder and the decoder.
展开▼