中文摘要：

针对城市交通安全和路口通行效率等问题,研究一种交通信号灯检测的技术。采用亮度分析、图像分割和形态学滤波对采集的图像进行预处理,排除背景干扰;利用RGB颜色空间下的颜色各通道差值分布检测交通灯颜色;最后基于对图像[0°,180°]的Radon变换找出峰值对应的角度,对图像在该角度上分别进行变换,利用形状特征对交通灯进行形状检测。采集自然环境下图像进行试验。结果表明,该算法的正确识别率达到90%以上,是一种较好的交通信号灯检测方法。

英文摘要：

A traffic lights detection technology is studied to solve the problems of urban traffic safety and crossroad traffic efficiency. The brightness analysis, image segmentation and morphological filtering are used to preprocess the acquired image to eliminate the background interference. The three-channel difference distribution of the three traffic lights＇ colors in RGB color space is adopted to detect the color of traffic lights. On the basis of the Radon transform for the image in 0°-180°, the angle corresponding to the peak value is found out, on which the image is transformed. The shape feature is employed to detect the shape of the traffic lights. The image acquired in the natural environmental was tested. The results show that the correct recognition rate of the algorithm is higher than 90%, which is a good method for traffic light detection.