Research On License Plate Recognition Algorithm The Smog Weather

In the smog weather,due to the scattering of the atmosphere,the reflected light diminishes.And because the reflected light of the atmosphere could participate in the imaging,therefore in the foggy day the image captured by the outdoor camera may result in produce-color-distortion,blurring,and image shifting during license plate recognition.Under this circumstance,the recognition rate of the system is significantly low,and cannot meet the standard requirements of engineering application.To solve the problem of low license plate recognition rate in haze weather,this paper proposes a license plate recognition algorithm aiming at haze weather problem fixing.This paper is arranged as following three parts:Pre-processing stage:First,it analyzes and summarizes the imaging characteristics by applying various effective algorithms.Some shortcomings of existing image dehazing algorithms such as details missing after image restoration,poor real-time performance,and time consuming are the reasons of these algorithms to be less practical.Based on the Retinex algorithm,this paper utilizes the Markov model to perform real-time image de-fogging to recognize the license plate with improved recognition rate.License plate positioning implementation:Based on the previous step,the obtained license plate image is positioned.In this stage,this paper optimizes license plate locating algorithm in complex environment by improving the accuracy of positioning.First,the Canny edge detection operator combines morphology and iterative fusion processing to obtain coarse license plate location,and then uses the improved Alex Net convolutional neural network algorithm to remove the fake license plate for precise positioning of the license plate.This solution solves the shortcomings of the current license plate location technology which has low positioning accuracy and takes a long time to process while in a complex environment.License plate recognition:In order to increase the recognition rate of license plate in haze weather,this paper improves the character recognition algorithm of convolutional neural network,and simplifies the network complexity through parameter adjustment and network layer design.This method can recognize the recognition time meanwhile still ensure the recognition accuracy,and reduce the complexity of the project.