雾霾天气下交通标志的检测与识别

Traffic sign detection and recognition under haze weather

针对我国当前交通环境的多变性,尤其是雾霾天气的出现,提出了一种在雾霾天气下快速实现交通标志分类识别的算法.首先,将图像从RGB颜色空间转换到HSV空间,对S通道图像边缘检测,计算标志牌的圆形度、矩形度等,判定标志牌的形状,定位提取标志牌所在的区域,根据标志牌的颜色和形状的匹配唯一性,对标志牌进行初步分类,然后提取标志牌所在的区域进行模板匹配,读取标志牌的具体含义.实验结果表明:该算法能有效地提高雾霾等恶劣天气下交通标志的检测率,并能满足实时性需求.

A fast traffic sign detection algorithm under haze environment is proposed for the characteristics of Chinese traffic environment variability, especially haze weather. First,transfer the image from RGB color space to HSV color space, edge detection with the S channel image, calculate signs of circle and rectangle, judging the shape of traf- fic signs and positioning its area. According to the sign of color and shape matching and uniqueness, preliminary classification is carried out on the sign, and then the sign area is extracted to template matching,read out the specific meaning of the sign. The experimental results show that the algorithm can effectively improve the rates of detection of traffic signs under haze weather, it also can satisfy the real-time requirements.