交通监控系统中基于多源信息融合的运动阴影检测

针对目前运动提取算法常将运动阴影错误检测为前景目标的问题，提出了一种交通监控系统中基于多源信息融合的运动阴影检测算法．在颜色空间中利用亮度、色度和边缘信息检测阴影的可能区域，使用逻辑“与”操作融合多源检测结果得到最终的阴影区域．与其他算法相比，新算法能更好地区分前景尤其是暗色前景及其阴影，提高了阴影区域的检测精度，更准确地实现了运动目标的提取．仿真实验表明，对不同颜色和尺寸的前景引起的阴影，算法都能鲁棒地分离目标及其阴影区域，前景提取效果好．

Moving Shadow Detection Based on Multi-Information Fusion in Traffic Monitoring System

Aiming at the problem that shadows cast by moving objects in traffic surveillance system were detected incorrectly as a foreground target by all most of the current moving detection algorithm, a method of moving shadow detection based on multi-information fusion was proposed, which can effectively detect the cast shadow of a moving vehicle in traffic image sequence. Luminance information, chrominance information and edge geometry information were used respectively to detect the candidate shadow regions. Finally, an logic AND operation was performed on the above candidate shadow regions in order to find the final shadow regions. Compared to other ways, the proposed algorithm has better results including the foreground segmentation from the moving shadow and detection precision, so the foreground can be extracted successfully. The algorithm has been tested under different conditions. The simulation results show that for the shadows caused by foreground with different color and size, this method can discriminate the target and shadow region effectively and robustly.