研究期間：10108~10207;With the prevalence of vehicular networks and cloud computing services, the next generation self-diagnostic and green energy vehicular transportation system integrate emerging communication systems, Vehicular Ad Hoc NETworks (VANET), sensor systems, intelligent video analysis systems, and cloud computing services to achieve enhancing traffic efficiency, congestion control, and improving traffic safety. In addition, better traffic conditions contribute to energy saving, improving air pollution, and environment protection. The main purpose of this sub-project in the integrated project is to provide useful information according to various and environment conditions and different user groups through automatic intelligent diagnosis and surveillance video analysis. Therefore, it plays an important role in the next generation self-diagnostic and green energy vehicular transportation system. In this project, we divide the system modules into three main parts and try to achieve them in three years. The goal of the first year includes daytime and nighttime basic traffic parameter extraction, enhanced vehicle tracking, and robust event analysis and detection. The content of the second year includes self diagnosis of lighting and weather conditions, and congestion level analysis and prediction for rainy days. The self-diagnosis ability should not be restricted to specified settings. And the congestion level analysis is performed based on constructing different training models for different surveillance scenes. The goal of the third year is to perform wide area surveillance by setting the cameras on high buildings. We plan to perform enhanced vehicle detection that is robust to noises and camera vibration, and perform flow analysis via regression analysis. In additions, the regions of interest in the scene should be extracted automatically via trajectory clustering.
