P. Wang

Pages: 103-112

Abstract
Lane detection is a key link in intelligent assisted driving systems. In order to improve the accuracy and real-time performance of lane line detection, the paper proposes an new lane detection method of intelligent assisted driving system based on convolutional neural networks. Firstly, along the scanning line of the LiDAR, a sliding window is used to extract road surface points. Secondly, based on the extraction results of road surface points, the edge of lane lines is enhanced through prior features of lane lines, and lane segments are extracted using line operation functions. Finally, normalize the extracted lane segments and establish a lane detection model through layer by layer training based on convolutional neural networks. The experimental results show that this method can improve the accuracy of lane detection, shorten detection time with maximum detection time 0.5 seconds.
Keywords: convolutional neural network; intelligent assisted driving vehicle; lane line detection; line operation function; normalization