Robust Lane Line Detection for Intelligent Vehicles under Complex Illumination Conditions Based on Image Processing

With the rapid advancement of intelligent driving technologies, visual navigation has become a core approach for environmental perception in autonomous vehicles.The accuracy of lane line detection under complex illumination directly limits the reliability of autonomous navigation.Existing methods face significant challenges: traditional threshold segmentation requires uniform lighting conditions; edge detection algorithms often produce false edges under strong light; deep learning methods suffer from high computational complexity and degraded feature extraction in low-light scenarios; and image enhancement techniques like histogram equalization (HE) struggle to adapt to dynamic lighting changes.To address these issues, this study proposes an integrated solution combining image enhancement and lane line detection.On one hand, an optimized Multi-Scale Retinex (MSR) method is employed to improve illumination component estimation and reflectance recovery, enhancing contrast in lane line images under complex lighting.On the other hand, a Sparrow Search Algorithm (SSA) is introduced to optimize the similarity matrix construction and cluster center initialization in spectral clustering, enabling precise separation of lane lines from the background.The proposed approach offers a robust and real-time solution for reliable navigation of intelligent vehicles in unstructured lighting environments, contributing significantly to the visual perception theory and advancing the industrialization of autonomous driving.