Complete description of multiple line segments using the Hough transform

摘要

The Hough transform (HT) is commonly used in machine vision applications for detecting discontinuous patterns in noisy images. The process of using the HT to detect lines in an image involves the computation of the HT for the entire image, accumulating votes in an accumulator array and searching the array for peaks which hold information of potential lines present in the input image. The peaks provide only the length of the normal to the line and the angle that the normal makes with the x-axis. They do not provide any information regarding the length, position or end points of the line segments. However, the butterfly shaped[1] spread of votes in the accumulator array, generated by the process of peak formation, holds vital information like the length and position of the input line segment. Some authors[2] have used this property to develop an algorithm to determine the coordinates of the end points, the length, and the normal parameters of straight lines. A limitation of this method, making it unsuitable for application to a real machine vision problem, is that it would yield erroneous results if applied to an image consisting of anything more than a single line segment. Moreover, the precision of this algorithm is dependent on the sharpness of the peak. In this paper, new techniques which address the above mentioned shortcomings have been described. This paper details the method developed to provide complete line segment description for an image consisting of multiple line segments. In addition, the developed techniques are more robust and accurate than the previously proposed methods as they do not depend upon the sharpness of the peak.