High‐speed line detection method using hough transform in local area
Kazuhito Murakami, Tadashi Naruse · Systems and Computers in Japan · 2001
Abstract In this paper, a high‐speed line detection method using Hough transform is proposed. This approach is inspired by features of human foveal vision. In particular, line segment candidates are extracted at sufficient level of accuracy by applying Hough transform inside a local window. After that, greater areas are generated by extension of line segment candidates, and segment verification is executed. Such local window is then moved sequentially until the entire image is covered. Such processing makes possible absorption of quantization errors of Hough transform as well as extraction of shorter segments that were hard to detect by means of standard Hough transform. Generally, the proposed algorithm supports the same level of accuracy as standard Hough transform while featuring acceleration (and computing cost reduction); moreover, further performance boost can be achieved through parallel processing. © 2001 Scripta Technica, Syst Comp Jpn, 32(10): 22–30, 2001