A windowing approach to detecting line segments using Hough transform

摘要

A new approach for using the Hough transform to detect line segments is presented. This approach is efficient in both space and time. Strategies combining the features of the intersection point [Ben-Tzvi, Leavers and Sandler, Proc. 5th Intl. Conf. Image Anal. 152–159 1990); Xu, Oja and Kultannen, Pattern Recognition Lett. 11, 331–338 (1990)] and dual plane [Conker, Comput. Vis. Graphics Image Process. 43, 115–132 (1988)] methods are used to calculate the Hough transform. A dense set of small overlapping windows are used to restrict the pairs of inage pixels that are evaluated. Experimental results indicate that this method reduces the time and space requirements significantly.