Components Recognition by the Generalized Hough Transform Using Multiple Two-dimensional Parameter Spaces

Abstract

The generalized Hough transform is a method for extracting similar figures to the template figure in an image. The generalized Hough transform needs huge memory for the fourdimensional parameter space which represents positions, scales and rotation angles of objective figures and requires processor power for conversing addresses from an image space into the parameter space. The paper proposes a method using multiple two dimensional parameter spaces for reducing the volume of the parameter space and for performing high speed processing. As the experimental result, the method provided high rate repeatability on the position, the scale and the rotation angle of extracted figures and performed short processing time.