Region-of-Activity Based Coding for Transform Domain DVC

Distributed video coding (DVC) has been widely acclaimed as an emerging video coding technology with immense potential for use in video surveillance systems due to its flexibility for designing extremely low complex video encoders. In this paper, we propose a novel modified framework for the DVC codec considering a region-of-activity based coding algorithm. The frames will be split into a sequence of regions based on the activity level that reflect the error distribution in the side information. The macro blocks will be re-ordered to concatenate the unevenly distributed errors and form new blocks for turbo coding to improve the efficiency of the puncturing process. The performance of the proposed technique is verified for three standard test video sequences and up to 21% bit rate saving is demonstrated.