The Method And Technology Of License Plate Recognition System With Haze Removal

The existing traffic intersection intelligent monitoring system does not have a good response to the effects of bad weather such as haze,which makes the intelligent monitoring system of traffic intersections unable to detect vehicles in a good condition in the event of severe weather such as haze.And identify the license plate.Moreover,the intelligent monitoring system lacks a complete framework for illegal vehicle detection,abnormal behavior identification,and license plate recognition,and the overall detection accuracy needs to be improved.In view of the problems existing in the license plate recognition system under haze and inclement weather,the focus has been placed on the removal of fog and the license plate recognition for vehicle violation detection,and a license plate system under smog weather has been constructed to effectively deal with hazy weather.Firstly,in light of the actual application requirements,the research significance and research background for haze removal and license plate recognition were explored,the status quo of removing haze and license plate recognition at home and abroad was analyzed,relevant technologies were analyzed,and the current domestic and international removal of smog and license plate recognition existed.To solve the problems summarized,formulate relevant research contents and analyze the relevant technologies applied in the process of implementing relevant research contents.Secondly,the technical features of haze removal are summarized,the atmospheric scattering model is analyzed,and the key factors for removing haze are determined.Based on the deep learning theory,the mapping relationship between convolution layer,pooled layer and non-linear regression layer is determined and used for construction.In order to remove haze from deep convolution neural networks and construct training sample sets and objective functions,a method for haze removal is proposed.Thirdly,the key steps in license plate recognition for vehicle violation detection are designed.The technical principle is described.The vehicle target detection is implemented based on deep convolution neural network,and the target tracking is performed based on the Kalman filter algorithm for the detected vehicle target.The idea to realize vehicle violation behavior identification.For the identified illegal vehicles,the license plate recognition work is achieved through the two-step vehicle license plate positioning method and the joint license plate recognition method of character segmentation and character recognition,thereby completing the final license plate recognition task that is integrated into the vehicle violation detection.Finally,the relevant methods and techniques for removing fog and license plate recognition were experimentally verified.The effect of deep convolution neural network layer on the removal of smog was tested and compared with the existing methods for removing haze.A database for vehicle detection and license plate recognition was established,a license plate recognition process that incorporated vehicle violation monitoring was implemented,and the real-time and accuracy of vehicle detection is verified.