Research On License Plate Number Recognition In Fog And Haze Weather

In recent years,with the progress and development of science and technology,license plate image recognition technology has been widely used.But more and more haze weather appears in our country,especially in northern cities,how to achieve accurate license plate recognition under haze weather has become a very important research direction.Under this background,this paper takes the license plate image recognition under the haze weather as the research direction.According to the research experience of many researchers,based on the Retinex theory,this paper studies the haze image enhancement processing method,and innovatively proposes a fusion algorithm combining the improved Retinex algorithm and the dark channel prior defogging algorithm to improve the image enhancement effect.The common location methods,morphology,projection and region merging are used.In the preprocessing method,the methods of license plate enhancement,binarization and edge detection are studied.Secondly,this paper introduces several commonly used license plate character segmentation algorithms,analyzes the characteristics of various algorithms and corrects the license plate,removes the license plate frame,segments the characters and normalizes the license plate characters,and obtains the segmented image.Finally,the operation of license plate character recognition module based on BP neural network,and get good results.