基于链码趋势表的交通灯形状识别方法

摘要

An algorithm of detecting the shape of a traffic light in a scene image and the concept of trend table is proposed.It tiltcer and enhances color information im RCB color space by the color fcather of traffie lingt.and gets the candidate areas.sements image using contour tracking akgirutgn abd figures out the chain code.It calculates the matching probaility by combining the symmetrical feather and length ratio feather,The discemible scale is determincd by the statistics.Expcrinental results using videos verify the effectiveness of the proposed algorithm.%设计一种交通灯箭头形状识别方法,提出链码趋势表的概念.根据交通灯颜色特征,在RGB颜色空间内进行颜色过滤和颜色加强,得到交通灯候选区域;利用轮廓跟踪算法分割图像,得到目标轮廓链码;计算链码在链码趋势表中的匹配率,结合箭头的对称特征和边长度比例特征得到交通灯箭头匹配率,并根据实验数据确定识别区间.实验结果表明,该方法的识别准确率较高.