中文摘要：

交通标志检测是影响交通标志识别速度和正确率的重要因素。采用一种 HSI自适应阈值粗分割方法获取场景图像的红色二值图，克服固定阈值分割算法受光照等外界因素影响较大的缺点。然后进行连通区域标记，排除了部分非禁令标志红色区域，并对多个相连禁令标志作有效分割，提取出禁令标志的候选区域。最后，通过分析候选区域外边缘的圆形相似度参数，确定了禁令标志区域。复杂道路环境下的场景实验表明：本方法具有较高的鲁棒性。

英文摘要：

T raffic sign detection is a key factor affecting the speed and accuracy of traffic sign auto-matic recognition .Fixed threshold segmentation result was easily influenced by external factors such as change of light ,so an adaptive threshold segmentation method based on HSI color space was firstly used to overcome the drawback .Then regions except some non-sign regions was la-beled ,regions with multiple connected signs was segmented and candidate regions were extrac-ted .Lastly ,the prohibition signs regions were determined through the analysis of the circle simi-larity parameter of external edges in the candidate regions .High robustness is demonstrated by simulation experiment under complex road environment .