The Research On Traffic Sign Automatic Detection And Recognition Algorithms

Road traffic problem has been a subject of concern. Traffic signs, as part of theroad traffic, plays a key role in providing information, regulating the behavior of thedriver, ensuring the smooth flow and security of traffic. This paper overview theresearch status of traffic sign detection and recognition, and design the basic frameworkof the system. The main work of the full text is as follow.(1) For the noise and motion blur generated during image acquisition, this paperexperiment with different ways of filtering, and choose better median filtering andWiener filter for image pre-processing operations.(2) This paper combine the image segmentation based on RGB and HSV colorspace with image morphological operations and binary image logic operations, and theexperimental results show that the image detection on HSV color space has better effect.(3) To the shape-based traffic sign detection, using the method of random Houghtransform to detect circular sign, and using the contour approximation method based onDouglas-Peucker algorithm for rectangular, triangular and octagonal traffic signdetection.(4) For the traffic sign recognition, this paper combine the SURF featureextraction and LSH algorithm with the RANSAC algorithm. At last, the experimentalcomparison show that the above method can achieve better recognition effect.Finally, the full text is summarized and prospected.