A Hough Transform based line detection method utilizing improved voting scheme

Detecting lines from a digital image is an important step in many applications. The Hough Transform (HT) is a powerful tool for line extraction due to its global vision and robustness in noisy and degraded environment. Aiming at solving the problems associated with the HT: the heavy computational cost and considerable degeneration in performance, a new method utilizing improved voting scheme for the HT is proposed. By separating the edge pixels into clusters of approximately collinear pixels, linear regression is used to find the orientation of each cluster. Judged by the value of determination coefficient, clusters are chosen for voting directly or voting around its main orientation. Gaussian blur is used in peak detection for reducing adjacent peaks. Experimental results show efficiency of the proposed method in terms of detection rate, time and memory saving, and the robustness to spurious lines.