Framework of Surveillance Video Analysis and Transmission System using Background Modeling and MIMO-OFDM

In this paper, a new video analysis and background modeling suitable for wireless video surveillance networks is proposed. Using video analysis, the proposed system can detect any location of important information from the video data, e. g., the location of persons, and then the whole video data can be drastically reduced by using background modeling except the above detected important information. In our experimental system, Motion JPEG video compression and 2x2-MIMO-OFDM wireless are used into a base system. The system can reduce of up to 87.41% of the video data using the proposed method with average PSNR of 25.92 dB using SD resolution image, with comparable image quality when background modeling is not used. Using HD resolution video simulations shows that the proposed schemes achieves slightly better PSNR performance.