Open AccessAn Efficient Lane Line Detection Method Based on Computer Vision
Author(s) -
Hongwei Zhu
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1802/3/032006
Subject(s) - preprocessor , computer science , artificial intelligence , process (computing) , computer vision , transformation (genetics) , line (geometry) , perspective (graphical) , artificial neural network , computation , image processing , image (mathematics) , pattern recognition (psychology) , algorithm , mathematics , biochemistry , chemistry , geometry , gene , operating system
Current approaches of using neural networks in lane departure warning systems are expensive. And it is difficult for neural networks to process 2K and 4K images. In this paper, we use a series of image preprocessing techniques such as perspective transformation, threshold processing and mask operation to process high-resolution images and the optimized sliding window method to fit the lane lines. Compared with neural network method, we can not only reduce hardware cost, but also quickly process high-resolution images. In addition, compared with the traditional lane line detection algorithm, we extract the region of interest through perspective transformation, which not only greatly reduces computation, but also converts images into an aerial view for subsequent processing. Especially, we carry out threshold operation and mask operation after perspective transformation, which greatly improves the performance of our algorithm in a strongly interfering environment. As can be seen from the experimental results, our method has good detection effect and can be applied to various road sections in different environments.