Lane Detection Based on Deep Learning and SSIM Method
Chao Ren, Xiuling Huang, Harutoshi Ogai
Abstract
Lane detection is an important part of autonomous driving techniques and is required to have high accuracy and robustness. However, due to the complicated change of weather and lighting, environmental effects such as fog, and the shape of the straight lane and curved lane, the application scenarios of lane detection are limited. To solve the above problems, we propose a novel lane detection method using deep learning and SSIM method to aim at challenging scenarios. The proposed method can detect lane using two deep learning detection methods in parallel. Then using the structural similarity index measure (SSIM) image similarity detection method to compare with the labeled actual lanes from ground truth and select the more accurate result as the output. Experiments showed that the lane recognition rate is high, and the speed is fast in various complex scenarios. The proposed method can improve the accuracy and robustness of lane detection.