论文部分内容阅读
路标识别方法是室外智能移动机器人机器视觉导航的关键技术之一。该文针对公路上的３种交通标志（箭头）提出了快速识别算法。算法通过搜索候选者、“第一次生长”、阈值自动选择、“第二次生长”４步从图象中分割出目标，而后进行分类。在基于轮廓结构分析的分类器设计中，利用“最短光栅距离”设计了直线判别准则，并用回溯法通过轮廓关键点的匹配来完成模式识别。整个算法结构简单，时间复杂度低，已成功地应用于ＴＨＭＲ－Ⅲ在校园道路上的实时导航，实验表明算法具有良好的实时性和鲁棒性。
Signpost identification method is one of the key technologies of machine vision navigation for outdoor intelligent mobile robots. This paper presents a fast identification algorithm for three traffic signs (arrows) on the highway. The algorithm divides the target from the image by searching for candidates, “first growth”, automatic threshold selection, and “second growth” in four steps, and then classifies them. In the classifier design based on the contour structure analysis, the criterion of straight lines is designed by using the “shortest grating distance”, and the pattern recognition is accomplished by the matching of the key points by the backtracking method. The whole algorithm has the advantages of simple structure and low time complexity. It has been successfully applied to real-time navigation of THMR-Ⅲ on campus. Experimental results show that the algorithm has good real-time and robustness.