Real-time Lane Detection Method Based On Region Of Interest

Aiming at the low accuracy and real-time performance of existing lane line recognition algorithms, a new lane line detection method based on region of interest was proposed. In the image preprocessing, nonlinear transformation is used to enhance the image contrast, and an improved multi-angle edge detection method is used to extract lane edges. According to prior knowledge, the recognition region of interest is established in polar coordinates, and the line segment is detected by the improved Hough transform method. The least square method is used to fit the dashed lane lines, and then the lane line direction is predicted according to the road model. The lane line identification is completed with Visual Studio 2013 platform. The experimental result shows that the algorithm can accurately identify lane lines and predict lane direction under variable lighting conditions, with the recognition accuracy up to 97.54%, and the average time consumption per frame is 0.015s. Compared with traditional detection methods, the accuracy of the proposed algorithm is improved by 9.1% and the average recognition speed is improved by 81.9%.