Enhancing license plate recognition in low-light conditions using Retinex_FM and YOLO model

The objective of this study is to address the challenge of reduced license plate recognition rates and accuracy under low-light conditions. To this end, a novel approach is proposed, integrating Retinex_FM (Retinex-based Fusion Method) image enhancement technology with YOLOv5s and YOLOv8m object detection models. Initially, license plate images captured under low-light conditions are processed using Retinex_FM to enhance the brightness and contrast of the images. These enhanced images are subsequently input into the YOLOv5s or YOLOv8m models for license plate detection. Experimental results demonstrate that Retinex_FM processing significantly enhances the license plate recognition performance of both YOLOv5s and YOLOv8m models under low-light conditions. The findings indicate that the Retinex_FM technique substantially improves the performance of object detection models in complex environments, providing a robust reference point for future research in license plate recognition technology.