摘要
针对色盲或色弱驾驶员特殊群体,结合交叉口无人驾驶系统,提出一种能识别出交通信号灯当前状态的方法。在HIS颜色空间进行交通灯颜色分割,利用交通灯被黑色矩形框包围这一典型特征进行形状分割,根据形状分割所得位置对颜色分割候选区域进行确认,从而精确定位交通灯位置以及亮灯在交通灯中的位置;同样根据被黑色矩形框包围的特征设计模板,最终利用模板匹配,完成交通灯的识别。实验表明,该算法识别准确率较高。
In this paper,a method of recognizing traffic lights through image processing is proposed.The system can help the visual impaired people driving through the signalized intersections.First,certain regions in an image are selected as color candidates from HIS color space.Then,based on the typical shape of traffic lights-black rectangular box,the algorithm will segment the image and select shape candidates.According to the position of candidate color segments and candidate shape segments,the position of traffic lights and the highlighted signal can be located precisely.In addition,with the template matching,traffic lights can also be recognized.The experiment results show that the proposed algorithm is stable and reliable.
出处
《交通信息与安全》
2011年第3期51-54,共4页
Journal of Transport Information and Safety
基金
国家自然科学基金项目(批准号:70871092)资助
关键词
交通灯识别
HIS颜色空间
形状分割
模板匹配
traffic lights recognition
HIS color space
shape segmentation
template matching
