Detection and recognition of road traffic signs establish a crucial element in Advanced Driver Assistance Systems (ADAS), which can provide real-time road sign information to vehicles. The purpose of conducting research on this topic is introduced to a less complex algorithm that works for traffic signs detection in complicated environment, accurately and rapidly. Initially, according to the color features of traffic signs, a color enhancement algorithm based on linear contrast stretching is used to enhance RGB images, and the color enhanced gray images of each channel are obtained. And then the dynamic thresholds are set according to pixel values of the obtained color enhancement gray maps to segment the images to obtain traffic sign candidate regions. Moreover, quite a few background interference is removed by morphological operation. Furthermore, the histogram of oriented gradient (HOG) features of candidate regions are extracted and SVM classifier is trained to accurately locate the candidate regions to further improve the detection accuracy. We performed some comprehensive experiments on the German Traffic Sign Detection Benchmark (GTSDB) dataset. The accuracy of traffic sign detection exceeded 97.41%. The proposed method has higher detection accuracy and time efficiency than other methods and better robustness under complex traffic environment.