Lane line robust detection and extraction method based on road constraint

• Main Authors: KATSURAHIKO, CHEN WANG
• Format: Patent
• Language: Chinese; English
• Published: 25.07.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention provides a lane line robust detection and extraction method based on road constraint aiming at the problems of abrasion, shielding and the like in lane line detection. The method comprises the steps that firstly, a large number of scene images of the Dongfeng park are collected, a lane line data set is constructed, and a road area is marked while lane lines are marked; and secondly, improving an ENet-SAD network model, defining a lane line segmentation loss function combined with road constraint, and generating an accurate lane line segmentation prediction result. And finally, a full-connection conditional random field algorithm is introduced to carry out post-processing on the output lane line probability graph, the probability graph is corrected, N off-line coordinate points are selected on each lane line, and a cubic spline curve fitting algorithm is adopted to obtain a fitting lane line. Experimental results show that compared with an ENet-SAD method, the method provided by the invention ach