基于交通标志都具有一定的颜色和形状，以红色倒三角形交通标志为例，介绍一种新的智能检测方法。该方法主要由以下4个环节构成。首先是颜色提取，将图像转换到HSI颜色模型的色调子空间，提取特定颜色——红色；其次是边缘检测，使用标准LOG模板在提取出来的红色区域上进行；这样，计算量将大为减少；然后是顶点判别，对边缘上的这些点，采用神经网络分类器进行分类，并分别对三角形的3种不同顶点作标记；最后是三角形的定位，通过一定的匹配准则提取出三角形。对不同场景下的20幅图像进行测试，检测正确率达到了100％。实验结果表明，该方法具有鲁棒性好、速度快、检测准确率高等特点。

This paper presents an intelligent method for detection of traffic signs. Which uses a particular color and shape for the detection of traffic signs. Red color and down triangle shape traffic sign is taken as an example, explain this method. This method is mainly carried out in four steps. First, convert RGB color space to HIS color space, and extract pixels with red color. Then perform LOG mask operation on the pixels got from step 1, for the detection of edges. By using neural network, we determine the angle pixels, and at the same time, we also determine on which specific angle the pixel is. And finally we detect the traffic sign by using the information of shape. We used 20 different images from different scenes to test this method, and the percentage of correctness is 100%. The experimental results reveals the good robustness, high speed , accurate detection rate of the presented method.