Tracking of objects containing multiple concentric arcs in a video stream, optimized with Fast Hough Transform

The paper proposes a method for tracking of objects containing multiple concentric arcs in a video stream, based on structure tensor analysis and a voting scheme in the arcs centers space. Proposed method was used to implement a wheel tracking system for the Automatic Vehicle Classification System. Algorithm is based on an assumption of axial symmetry of the object. The algorithm is optimized with Fast Hough Transform. Numerical experiments are presented to evidence the efficiency of the method.