Research On Denoising And Character Recognition Algorithm In License Plate Recognition

In the 21st century,where technology is rapidly developing,every household is enjoying the benefits of various means of transportation.The integration of technology into life has brought us convenience,but it has also made transportation management more and more difficult.Therefore,in order to solve urban traffic management,the state has proposed an intelligent traffic management system.As the core technology of intelligent transportation system,license plate recognition system brings great convenience to urban traffic management and national travel.The exploration of various aspects of its technology has attracted the enthusiasm and attention of more and more researchers.This paper focuses on how to improve the work efficiency of the license plate recognition system,starting from improving the image quality of the license plate and improving the efficiency of the recognition algorithm.The main work of the thesis is as follows:1.In order to improve the quality of license plate image,this paper studies the traditional non-local mean denoising algorithm,and finds that its measurement similarity is unstable and easily leads to edge detail loss.This paper proposes a license plate image denoising algorithm based on edge detection.The algorithm uses the standardized Euclidean distance instead of the simple Euclidean distance to measure the similarity between neighborhood blocks,and uses the improved Canny edge detection operator for edge independent denoising.By comparing with the classical algorithm,it is proved that the improved algorithm improves the accuracy and stability of the similarity measure,and better preserves the edge detail information.2.In order to further improve the efficiency of license plate character recognition,this paper studies and analyzes the template matching algorithm and convolutional neural network.It is found that the template matching method has low accuracy in identifying similar characters,while the convolutional neural network has low recognition time efficiency.In response to the above shortcomings,a new character recognition algorithm is proposed.The algorithm uses template matching to judge similar characters,and uses a modified convolutional neural network to perform secondary recognition processing on similar characters.The experimental results show that the character recognition rate of the algorithm is as high as 98.6%,and the recognition time is only 35.5 milliseconds.It ensures the high recognition rate and significantly reduces the recognition time,which has certain practical and theoretical guiding significance.Figure[32]Table[8]reference[57]...