自然场景车载视频道路交通限速标志的检测与识别方法

A Detection and Recognition Method for Traffic Speed Limit Signs Based on Vehicle Videos

根据交通限速标志的颜色和形状特征,提出一种基于车载视频的交通限速标志的检测与识别方法。首先,将原始影像转换到色调、饱和度和明度空间,利用限速标志红色圆环颜色特征设置红色分割阈值,去除大片非感兴趣区域;然后,根据限速标志的圆形特征来检测限速标志,对检测到的限速标志进行字符分割,提取分割字符分块形成的特征向量,通过特征匹配识别出限速标志。结果表明,设计的方法比模板匹配法识别正确率高,可有效对自然场景中的限速标志进行检测与识别。

An approach was proposed to detect and recognize traffic speed limit signs of vehicle videos according to the color and geometric features of the signs.Firstly,the detected image is transformed into HSV space,the color features of the speed limit signs are used to remove the uninteresting regions,Then the speed limit signs will be detected by its circular features.A recognition method based on character segmentation is realized.Experimental results show that the proposed method is more effective for recognizing traffic speed limit signs based on vehicle videos and has a higher recognition rate than pattern match method.