A Novel Method for Wheel Rim Recognition

Guan Xu, Jian Su, Hongda Pan, Donglin Zhang

Abstract: The recognition of the wheel rim is a crucial area for vehicle safety. We present a novel recognition method for the detection of wheel rim whose projective curve is an ellipse. An image preprocessing method including the Otsu algorithm to binary the gray image, removing spurious areas and region filling algorithm is introduced to segment the rim area correctly. Then the ellipse rim region is considered as a 2D sheet and the centroid coordinate of the ellipse region is identified by the formula of calculating mass center. At last, all the five parameters to construct the ellipse of wheel rim are obtained according to the geometrical property that the optimum vertex on the semiaxis is the point on the edge of ellipse whose distance between the centroid and the point is maximum or minimum. The experimental results show that the proposed method recognizes the wheel rim efficiently and precisely.