Hough transform using slope and curvature as local properties to detect arbitrary 2D shapes

Abstract

A Hough transform algorithm is presented with a complexity of o(mn) instead of o(q/sup 2/mn) or o(qm) as in the Marlin-Faber algorithm or Ballard algorithm. (m and n are the numbers of points on the model's and shape's boundary, respectively, and q is the quantization of the parameter space). This method has been used to detect arbitrary shapes and for recognizing partially occluded parts. It is noted that although the time complexity of the Hough transform is reduced, more computations are needed in image space.<>