基于RGB归一化模型与分群算法的交通灯检测

摘要

交通灯检测是智能交通系统研究的一个重要方向，采用了一种交通灯实时检测算法。首先，对图像进行RGB归一化处理，得到了几种颜色独立的图像。然后提取图像的二值化信息，并作形态学处理。接着对图像进行分群操作和标记处理，并且根据交通灯形状区域特点做出边长与面积条件判断。最后通过提取交通灯的样本空间来进行匹配，从而锁定交通灯位置。实验部分将其与传统的 Hough 算法进行比较，结果表明所提算法可以更快且准确地检测到交通灯。%Traffic lights detection is an important research direction of intelligent transportation system with adopts the real-time detection algorithm adopted.Firstly,the image is processed by RGB normali-zation model,which produces several independent image colors.Secondly,the algorithm extracts the bina-rization information from the image and makes the morphological processing.Thirdly,it exerts the clus-tering operation and labeling processing for each image and makes the condition j udgment on the length and area according to the shape characteristics of traffic lights region.Finally,the value is extracted from the sample space of traffic lights and matched with the one from the real-time scene.Therefore,the posi-tion of traffic lights is locked well.The experiment compares the Clustering algorithm with the tradition-al Hough algorithm and the results show that the proposed algorithm can quickly and accurately detect the traffic lights.