Fuzzy adaptive pre-processing models for road sign recognition

We proposed fuzzy inference schemes to address the changes of the lighting environment problems: the illumination of the images captured from camera installed on a moving vehicle also varies from frame to frame. First, the input image is checked with a fuzzy inference method to evaluate the illumination conditions in order to apply appropriate preprocessing operations to get a better result. To overcome the effects caused by vehicle speed and changes in direction, a fuzzy inference method was again used to select an adapted detection window to increase the throughput rate. The Adaboost classifier was employed to detect the road sign candidates from an image and the support vector machine technique was employed to recognize the content of the detected road sign. The mandatory and warning road traffic signs are the processing targets in this research. The proposed system can detect and recognize road signs correctly from the captured image, and not only overcome problems such as low illumination, viewpoint rotation, partial occlusion and rich red color around the road sign, but also reach a high recognition rate and processing performance.