Abstract: Focused on the problem of time consumption and accuracy of traffic sign recognition system, an efficient algorithm for traffic sign detection based on improved color detection in RGB color space combined with support vector machine is presented. Firstly, the histogram equalization and Gabor filter are applied for image enhancement to highlight the image color. Secondly, an improved color detection method is proposed to select the coarse candidate areas. Thirdly, a SVM classifier is trained using HOG feature for further detection and shape judgment. Comperative tests are done about detection accuracy and detection time consumption. The experimental results show that this algorithm testing time is short, error detection rate and false detection rate are low. The algorithm can detect signs quickly and determine the shape. It can also deal with various cases such as images with low brightness, rotation and partial occlusion with high accuracy. Compared with color enhancement algorithm, this method can achieve better accuracy and has less time consumption.