Enhancing License Plate Recognition in Foggy Conditions Based on YOLOV5 and MOSAIC Method

Abstract

The high precision recognition accuracy of the license plate for driving system is a challenging issue, especially in low-light conditions, such as foggy days. This paper introduces a foggy object detection network based on YOLOV5 to extract the relevant features of license plate images and train them for detection tasks, and then introduces the MOSAIC enhancement algorithm to recognize license plates in real time. The whole recognition process is mainly divided into two processes: license plate positioning and license plate character recognition, which is conducive to reducing the accumulation of recognition errors and improving the efficiency of recognition. According to the test results, the detection accuracy of the enhanced YOLOV5 is significantly improved, and its ability to identify and manage traffic vehicles is improved, with accurate and effective vehicle information.