Rapid improvement in computing power, cheap sensing and more flexible algorithms are facilitating increased development of real-time video surveillance and monitoring systems. The development of video understanding systems in certain critical applications in the real world can be done only if performance guarantees can be provided for these systems and by performing a careful analysis of influence of various tuning parameters on the systems.; In this thesis, the performance characterizations of low-level vision algorithms including background maintenance, morphological filtering have been presented. The research results systematically help the system designers increase the robustness of the video surveillance system. |