Title :
A fast hough transform based on C3TR using shift operation
Author :
Kondo, Yuki ; Numada, Munetoshi ; Koshimizu, Hiroyasu
Author_Institution :
Sch. of Inf. Sci. & Technol., Chukyo Univ., Nagoya, Japan
Abstract :
The Hough transform is an important scheme for providing a robust pattern recognition due to the global potential for extracting lines from an image. In order to realize the faster Hough transform, it is the most essential to reduce the cost in both trigonometric function and multiplication calculations. This paper proposes a fast line detection method that generates an accurate Hough curve can be generated by a single operation of multiplication using Cosine-3-Term-Recurrence. Moreover, the high-speed computation method which uses shift operation instead of multiplication was also shown in this research. Doing experiments has confirmed that the proposed method provides the best accuracy and the fastest computation.
Keywords :
Hough transforms; object detection; C3TR; Hough curve; Hough transform; cosine-3-term-recurrence; line detection method; multiplication calculations; robust pattern recognition; shift operation; trigonometric function; Approximation error; Computational efficiency; Hardware; Image edge detection; Mathematical model; Transforms; C3TR; Hough Transform; Line Detection; Shift Operation;
Conference_Titel :
Frontiers of Computer Vision (FCV), 2015 21st Korea-Japan Joint Workshop on
Conference_Location :
Mokpo
DOI :
10.1109/FCV.2015.7103700