指示类交通标识的自动检测

Automatic detection of indicative traffic signs

针对指示类交通标识的特点,进行标志牌检测算法的研究.首先利用快速HSV颜色查找的方法进行颜色分割,对颜色分割后的二值图进行必要的形态学预处理,接着利用圆形度和拐角检测与几何特征结合的方法进行形状识别.针对某些圆形标志边缘分割不完整的情形,提出一种根据距离直方图判断是否为圆形标志的方法.实验结果表明,算法对实际道路环境下的指示类交通标识的检测具有较好准确性和鲁棒性,满足实时性要求.

This paper studied the detection algorithms of indicative traffic signs. First, the color segmentation was recorded along with a fast HSV color search method, and later the morphologic preprocessing was taken on the binary image after the color segmentation. Second, circularity and corner detection combined with geometric characteristics were used to judge if a region was an indicative sign. Last, in view of the incomplete edge of the round sign after the color segmentation, a new algorithm was raised based on a distance histogram for confirmation. Experimental results show that this algorithm is accurate and robust for detecting indicative traffic signs in real traffic environments, and also meets the real-time demand.