Authors: Hyunhee Park
Addresses: Department of Computer Software, Korean Bible University, Seoul, South Korea
Abstract: This study proposes a lane detection method based on expressway driving videos through a computer vision-based image processing system without using sensors. Both straight and curved sections can occur on a road, and thus, lanes must be detected by quickly determining such sections. The proposed method detects straight and curved sections that are estimated to be lanes using the Hough transform. When lanes are detected from actual images, the scope of left and right lanes is limited to reduce computational load. In this paper, we propose a lane-detection algorithm using the colour space and a stepwise algorithm for accurate lane detection. To verify the proposed algorithms, we developed a small self-driving vehicle model using a TX-2 board. The experiment results when applying the proposed Hough transform algorithm and lane-detection algorithm using the colour space show that the lane detection rate of vehicles driving on curves at high speed is approximately 96%. Through the extensive simulation results, the proposed algorithm to vehicle black boxes or autonomous driving will help prevent lane departure and reduce accident rates.
Keywords: lane detection; Hough transform; self-driving vehicle; OpenCV; TensorFlow.
International Journal of Web and Grid Services, 2019 Vol.15 No.3, pp.240 - 250
Received: 17 Jan 2019
Accepted: 07 Feb 2019
Published online: 04 Jul 2019 *