基于差分均值背景提取和矩阵分区目标检测算法的研究

为提高对车辆图像的检测程度和实时性,针对智能交通系统,通过对实时路况的信息采集和视频图像的处理,提出了一种基于差分均值的背景提取计算方法和矩阵分区域的阴影检测方法,最终得到一个视频车辆的检测原型,从而实现对运动车辆的检测.实验结果表明,此种方法简单、计算量小、鲁棒性高,能快速地提取出背景图像,检测出比较完整的车辆阴影,可满足多运动目标的实时检测要求.

Background Extraction Based on Differential Mean Method and Shadow Detection Using Matrix Subregion Partition

For intelligent transportation monitoring systems, a background extraction method based on differential mean and shadow detection using matrix subregion partition are proposed to improve the quality and speed of the vehicle image detection. The real-time traffic information was collected and its video images were sophisticatedly processed. Finally, a video vehicle monitoring prototype was setup to realize the detection of moving vehicles. Experimental results show that this method has the advantages of simplicity, robustness and small amount of calculation. It could extract a background image quickly and detect a complete shadow of vehicles. This method could meet the requirements of real-time detection of multiple moving targets.