License Plate Recognition System Based On Wavelet Analysis Research

With the background of economy growing and the quality of people’s lives improving continuously, the vehicle sale grows quickly along with the increasing demand for the convenience of vehicles. What’s more, traffic congestion are always been resulted in the limited road because of the ownership of car. Thus, all aspects of the vehicle regulation are facing major test.The emergence of intelligent transportation system demonstrates it’s importance on these thorny issues obviously. One of the key technologies is license plate recognition system, which is playing a huge role and exposes its tremendous economic value. Depth research of the license plate recognition system becomes more and more important. Considering that wavelet transform has good time-frequency analysis features in the field of image processing, therefore, license plate recognition has been studied by wavelet analysis in this paper.First, we studied a car license plate location algorithm based on wavelet analysis in the positioning part of the vehicle identification systems. Besides, a comparative analysis between the conventional gradient edge detection operator and wavelet transform decomposition of the edge has been done respectively in this vehicle image edge extraction. Then, through the preprocessing, morphological, multiple connectivity domain analysis technology integration, we can effectively realize the license plate location based on wavelet analysis, which can locate the license plate in many cases accurately. The experimental tests show that it has a higher positioning accuracy.Secondly, we take use of wavelet transform to denoise in the part of the plate segmentation. we can get the license plate tilt angle through the hough transform and radon transform and can even get a comparative analysis between of them. After a series of steps, such as:a rotation angle of inclination, removal of borders, segmented characters, accurate license plate characters with good usability and robustness are coming.Finally, research and analysis on the character feature extraction in the character recognition module have been done, we can use the wavelet transform decomposition coefficients of character image and the best base of wavelet packet as the feature vectors respectively. At the same time, we can enter a support vector machine to identify, which has a higher recognition rate. After that a comparative analysis of the performance between the two has been done through the experimental data. At last, we make it come ture that achieve the accurate identification of the vehicle images by the use of MATLAB.