Lane Detection Based on Least Square Fitting

Lane detection is a key technique for intelligent vehicle driving. Aiming at the detection performance of existing lane detection algorithms, based on least square fitting, we propose a lane detection algorithm for structural road. The lane videos are gotten by the monocular camera installed in the car. Image preprocessing is applied to improve image contrast and then the image is segmented by improved Otsu. At last, the current lanes are extracted and equations are rebuilt by the least square fitting. The experiment results show that the proposed method has better accuracy and robustness compared to existing lane detection algorithms.