Cross-layer optimized wireless video surveillance.

摘要

A wireless video surveillance system contains three major components, the video capture and preprocessing, the video compression and transmission over wireless sensor networks (WSNs), and the video analysis at the receiving end. The coordination of different components is important for improving the end-to-end video quality, especially under the communication resource constraint. Cross-layer control proves to be an efficient measure for optimal system configuration. In this dissertation, we address the problem of implementing cross-layer optimization in the wireless video surveillance system.;The thesis work is based on three research projects. In the first project, a single PTU (pan-tilt-unit) camera is used for video object tracking. The problem studied is how to improve the quality of the received video by jointly considering the coding and transmission process. The cross-layer controller determines the optimal coding and transmission parameters, according to the dynamic channel condition and the transmission delay. Multiple error concealment strategies are developed utilizing the special property of the PTU camera motion.;In the second project, the binocular PTU camera is adopted for video object tracking. The presented work studied the fast disparity estimation algorithm and the 3D video transcoding over the WSN for real-time applications. The disparity/depth information is estimated in a coarse-to-fine manner using both local and global methods. The transcoding is coordinated by the cross-layer controller based on the channel condition and the data rate constraint, in order to achieve the best view synthesis quality.;The third project is applied for multi-camera motion capture in remote healthcare monitoring. The challenge is the resource allocation for multiple video sequences. The presented cross-layer design incorporates the delay sensitive, content-aware video coding and transmission, and the adaptive video coding and transmission to ensure the optimal and balanced quality for the multi-view videos.;In these projects, interdisciplinary study is conducted to synergize the surveillance system under the cross-layer optimization framework. Experimental results demonstrate the efficiency of the proposed schemes. The challenges of cross-layer design in existing wireless video surveillance systems are also analyzed to enlighten the future work.