基于颜色和形状的道路交通标志检测  

Road Traffic Sign Detection Using Color and Shape

作　　者：陈维馨[1] 李翠华[1] 汪哲慎[1] 

机构地区：[1]厦门大学计算机科学系,福建厦门361005

出　　处：《厦门大学学报（自然科学版）》2007年第5期635-640,共6页Journal of Xiamen University：Natural Science

基　　金：国家自然科学基金(60175008);国家创新研究群体资助项目(60024301);厦门大学985二期信息创新平台项目资助

摘　　要：提出了一种基于颜色分割和形状分析的交通标志检测方法.首先利用改进后的简单向量滤波器分割感兴趣色彩区域,对不同色彩区域进行相应形状检测.采用拐角特征与几何结构分析相结合的方法检测矩形与三角形标志,提出了一种基于对称性特征的简单编码算法,并应用于圆形检测.实验结果表明,该方法能在多种复杂场景中有效地定位交通标志,对天气变化有一定鲁棒性,为后继的交通标志识别工作奠定了良好基础.Based on color segment and shape analysis, a method of road traffic sign detection is proposed. The method first uses simple vector filter （SVF） to extract certain colors. Different shape analyses are done in regions with distinct color features, and then uses corner extraction together with geometric feature analysis to detect rectangular signs and triangle signs. Circular signs are searched by a new kind of code based on the symmetric nature of these signs. Experiments show good results of detecting road traffic signs under complex backgrounds and different weather conditions. The method is good for further processing such as road traffic sign recognition.

关 键 词：交通标志检测 颜色分割 拐角提取 几何特征分析 圆形编码 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]