A lane line detection method based on adaptive ROI selection for smart networked vehicles

The assisted driving and autonomous driving of intelligent networked vehicles have become an inevitable trend in the development of the automotive industry, and the detection of lane lines provides the basic core function for intelligent driving technology, and the accuracy, real-time, and robustness of recognition directly affects the safety and reliability of the vehicle in road driving. This paper focuses on two aspects of image ROI extraction and lane line target detection for lane line detection and proposes a lane line detection method based on adaptive ROI selection for intelligent networked vehicles. Firstly, to address the problem of imprecise delineation due to the influence of road surroundings in ROI extraction for lane line detection, a dynamic ROI extraction method based on regional grayscale mutation is proposed to select more accurate ROI delineation utilizing regional delineation and weighted summation. Then for the problems of the OTSU binarization method, this paper combines the grayscale histogram features to extract the effective threshold to get the binarization map. Hough transform is the most commonly used method in lane line straight line detection; for the shortcomings of traditional Hough transform with high time complexity and poor accuracy, this paper proposes an improved lane line detection method based on hough transform, which can effectively improve the lane line detection effect. Finally, the accuracy and effectiveness of the method for lane line detection are shown through experimental results.