【摘要】 针对传统特征匹配车速检测方法实时性较差的问题,提出一种改进的角点特征匹配车速检测方法。基于视频图像,采用混合高斯模型检测方法提取运动车辆目标,利用Harris算法检测车辆目标的角点特征,将运动估计和NCC匹配相结合,优化匹配区域搜索方法,对车辆目标角点进行角点粗匹配,再通过RANSAC算法进行角点精匹配和单视测量坐标转换以实现车速检测。实验结果表明,与传统方法相比,该方法的角点粗匹配速度提高400%,角点精匹配速度提高200%,车速准确性达到90%以上,能有效提高车速检测的实时性和准确性,满足实际车速检测的要求。更多还原

【Abstract】 In order to improve the real-time performance of traditional feature matching vehicle speed detection,this paper presents an improving Harris corner matching method.Based on video,it uses moving object detection and Harris corner detection,combines motion estimation and NCC template matching,optimizes matching area search strategy for corner coarse matching,then uses random sample consensus for corner fine matching and single view metrology coordinate transformation,and finally achieves vehicle speed measurement.Experimental results show that,compared with the traditional method,corner coarse matching speed is increased by four hundred percent,corner fine matching speed is increased by two hundred percent,the vehicle speed accuracy reaches more than ninety percent,and the new method successfully improved the real-time of the algorithm and the accuracy of corner matching.更多还原