Research On Algorithms And Mathematic Model Of Road Sign Automatic Detection And Recognition

With the fast development of economy, our country enter the automobile society quickly, the traffic safety and smoothing are concerned widespread by all society. In this kind of condition, the people start the research the intelligent transportation system (ITS). The intelligent transportation system (ITS) includes the intelligent infrastructure and the intelligent vehicles, is a comprehensive research field, its core technique involves pattern recognition, image processing, digital signal processing, artificial intelligence, electronics technique, information technique, communication technique and the system engineering technique etc.The road sign recognition (RSR) the important component of intelligent vehicle, it collecting and recognition the road sign information in the vehicle driving process, then gave alert or warnings to driver, or control the operation of the vehicle directly, to keep the transportation smoothing and avoiding traffic accident; The automatic segmentation of road sign and recognition is the important support software of the intelligent transportation system, there are important theory and practical values.Our country's research on intelligence transportation system get to start just in recently, and the road sign recognition have so much difficulty, so at percent, there is no real system is put into application. But in the same time, the comparatively simpler license plates identify and container type identify or other recognition systems have already put into application and have some good results. Fortunately, some research institute and university in our county have start pay attention to, and begin to study the road traffic sign recognition, and give some initial results.The processing object of automatic road sign segmentation and recognition is a traffic sign image that has the complicated background, otherwise the light of the road sign, deterioration in color, transform, and distortion etc. problem. So in the RSR system, there are many key problem need to be study.In my thesis, the main contents of study include image segmentation based on color (first classification), the quickly shape analysis (second classification), the theories research of Laplace kernel statistical classifier (third classification). in the study of image segmentation based on color information, through study of shape and color characteristic of road sign, we propose a segmentation method based on HSV color model, this method ,the variety of lighting have few effect to color segmentation. In this color model, H (hue) denote different color, such as red, yellow, blue and etc. S (saturation) denote the depth of color, V (value) denote the quantity of lighting. This method has good adaptive character toward lighting, deterioration in color. Different kinds of road sign have different shape, so shape is another key characteristic, in the section of shape analysis; we propose a simple and fast method. Laplace kernel statistical classifier proposed in we thesis have great advantage compare with other methods. This classifier model have characteristic of self-learning and easily scaling, it is reasonable using it for road sign recognition.