The segmentation of ban traffic signs based on two-step randomized Hough transform

In order to segment ban traffic signs quickly and accurately, this paper first refers to its characteristic of red circle and white background to start color segmentation, then presents a segmentation method based two-step randomized Hough transform (RHT) segmentation method. The first step, extracting the approximate area in the ban traffic signs, which is using the means of Hough transform detect approximate circle method, then screening edge chain code by the approximate range of the center and radius of traffic signs. The second step, comprehensive the ideas, such as the tangent of edge chain, the ellipse polar and polar line, the distance of the point on the ellipse to the two elliptical focal point is equal to long-axis length, Using these ideas to improve the randomized Hough transform and segment the ellipse precise area where traffic signs at the original image. This method uses the idea of thin after rough, avoiding a large number of invalid samples. Experiments show the effectiveness of the new method.