Research on surface defect imaging detection and traffic safety optimization of vehicle-mounted lidar wedge prisms

Wedge prisms present significant technical challenges in optical detection due to their unique geometric structure. The acquisition of high-quality datasets remains a critical challenge in this field. This study proposes a novel method based on incident angle optimization and reflection area prediction to enhance imaging quality. The method ensures optimal component quality during manufacturing processes before LiDAR system integration. Using Mie scattering theory, we determined the optimal incident angle range of ${20}^{\circ }$–${40}^{\circ }$. Dual wavelength imaging demonstrated significant improvements across multiple performance indicators compared to single wavelength approaches. The established model relating wedge thickness to light reflection characteristics successfully predicted physical boundaries of two distinct reflection areas. Experimental validation confirmed the existence of interference reflection and geometric reflection mechanisms. The prediction error between the theoretical and actual image reflection areas remained within $20$ pixels. Quantitative analysis reveals that micron-level optical defects significantly impact vehicle mounted LiDAR performance. This impact becomes particularly pronounced under heavy rainfall conditions, substantially reducing system reliability. This framework provides a foundation for improving LiDAR manufacturing quality control and enables future development of automated inspection systems.