A Nonlinear Oscillator Coupling Model to Provide the Hough Transform Function Without the Discrete Voting Procedure

Abstract

The most common algorithm for detecting straight lines in a digital image is the Hough Transform method. The method involves two consecutive steps: transforming the image edge location onto parameter space and voting procedure to acquire line characteristics. The former deals in continuous space, and the latter counts votes in a discrete matrix, making the algorithm sensitive to noise. We transformed each image pixel location to the parameter space as a Hough line and generated a nonlinear oscillator to attract them into the intersection with other lines. This concept could contribute to a line detection problem in image processing using a consistent mathematical description.