雾霾天气下交通信号灯的识别

Identification of Traffic Lights in Haze Weather Conditions

针对雾霾天气下交通信号灯定位准确率较低、图像增强时出现图像亮度不均匀的问题,该文提出一种基于改进的带色彩恢复的多尺度视网膜增强(Multi-Scale Retinex with Color Restoration,MSRCR)的雾霾天气下信号灯识别算法。首先利用改进的MSRCR算法对有雾图像进行预处理,校正图像亮度并丰富图像细节;再利用最大稳定极值区域(Maximally Stable Extremal Regions,MSER)算法以及信号灯的背板信息确定信号灯的位置;最后将定位区域转换至HSV空间进行信号灯识别。结果表明,该方法能够在雾霾条件下有效地定位及识别交通信号灯。

In order to solve the problems of low location accuracy of traffic lights and uneven brightness during image enhancement in haze weather,an improved Multi-Scale Retinex with Color Restoration(MSRCR)weather signal recognition algorithm based on color restoration is proposed in this paper.Firstly,the improved MSRCR algorithm is used to preprocess the foggy image to correct the brightness of the image and enrich the image details;then the Maximally Stable Extremal Region(MSER)algorithm and the backplane information of the signal lamp are used to determine the position of the signal l finally,the location area is converted to the HSV space for signal light recognition.The results show that this method can effectively locate and identify traffic lights under haze conditions.