Investigation of lane line boundary tracking technology utilizing eight-neighborhood inverse perspective transformation algorithm

Lane line boundary detection algorithms primarily rely on camera-acquired image data for accurate recognition. However, during image acquisition, distortions caused by glass refraction and the inherent imaging principles of the camera can compromise the precision of lane line detection. To enhance the precision of road environment perception, this study first conducts a comprehensive camera calibration to determine the intrinsic and extrinsic parameters, which serve as the basis for subsequent geometric distortion correction of the acquired images. Subsequently, Otsu’s method is employed for binary segmentation of lane images, and Inverse Perspective Mapping (IPM) is utilized to transform the perspective and extract precise lane line boundary coordinates. The eight-neighborhood boundary tracking algorithm is then employed to track lane lines, improving image processing speed while reducing false positives and false negatives. Finally, the Least Squares Method is utilized to fit the extracted boundary points of the lane lines, thereby constructing an accurate mathematical model for lane representation. The experiment was conducted using multi-element lane lines as the test platform. The experimental results demonstrate that the proposed algorithm achieves nearly a fivefold improvement in detection speed compared to the method without inverse perspective transformation. The Intersection over Union (IoU) between the extracted lane lines and the ground truth reaches 0.88, approaching the ideal value of 1.0. The pixel-level accuracy of lane line extraction is 93.7%, and the recall rate, representing the proportion of correctly extracted lane line pixels, is 95.2%. Furthermore, the balanced F1-score, which evaluates the trade-off between precision and recall, reaches 0.98, indicating excellent overall performance. Additionally, it exhibited superior robustness. These findings suggest that this method can significantly enhance the accuracy and efficiency of lane line detection in complex road environments, thereby providing reliable technical support for intelligent driving and Advanced Driver Assistance Systems (ADAS).