Research on road recognition and initiative security system of computer vision-based

In this paper, the images of forward scenery acquired by CCD mounted on vehicle, will be segmented into road surface, lane mark and obstacle by means of detecting the road edge with the method of wavelet transformation. The method of Hough transformation is employed to obtain the coordinate of the road, and then to recognize the forward obstacle. Based on the time-model of safe steering and integrated the set of safe-time rules, a model of knowledge expression of steering safety-radix-feature state body is presented in this paper. If it's not satisfied with the restrict condition of dynamic safe-time, the system will present alarm, with reasoning the current steering sate. With the analysis and experiment, it has proved that the method presented in this paper is feasible.