车辆越界快速自适应检测方法研究

Research on Adaptive Method of Detect Vehicle-cross-boundary Rapidly

针对智能交通系统对快速性和准确性的要求,提出了基于区域生长的边缘关键点提取和基于HSI色彩空间的车辆违规越界自适应检测方法。首先选取车道线中心的点,以此为种子点进行区域生长,均匀选取区域边缘上的某些点作为关键点。然后将图像中的RGB表色模型转换为HSI表色模型,根据HSI表色模型中关键点的像素特征设定二值化阈值,这样能较好地克服光线变化对检测结果的影响,同时能够适应各种颜色车辆的检测,提高检测速度。

Focusing on the requirement for rapidity and accuracy of intelligent transportation systems,an adaptive method of detecting whether vehicles erosses the boundary based on HSI color space is presented,and within this method the important points on the edge is extracted based on region-grow- ing. Firstly,a point in the center of the traffic lane is chosen as the seed point to conduct region-growing,and some points on the edge are also chosen e- venly as the important points. Then,images are converted from RGB designation model to HSI designation model ,thresholds used in binary conversion are specified in terms of the pixel character of the important points. In this way,final results are less sensitive to the changing light. This method is efficient to detect vehicle in different colors and enhance the detection velocity.