Research Of License Plate Recognition Technology Algotithm Based On PCA-SIFT

With the rapid development of our country’s economy, the requisition on the traffic control and regulate of the intelligent traffic system improves day after day. Vehicle License Plate Recognition System(LPRS)using computer do processing, analysis and recognition for video traffic images to extract plate information. Objective is to facilitate management, such as traffic control, statistics, and fees. And LPRS is one of the hot core issues of intelligent traffic management. Many scholars at home and abroad are specializing in license plate localization and character recognition technologies, and have achieved many results, but there are still many deserve further research and improvement. For example, when the complex environmental change, these algorithms is powerless due to its narrow limitation. David g. Lowe put forward the improved SIFT feature extraction algorithm on the basis of predecessors’ work, in2004[1-2]. SIFT algorithm scheme to reduce the dimension of the extracted feature points in the original algorithm, so as to make the license plate recognition rapid and accurate. When facing the complicated situation, such as rotating background, shade scale zooming, outside noise and so on, SIFT algorithm does well in practice. But in the description of feature points, SIFT algorithm spends too much time on computation due to using the higher dimension and the algorithm extracts too many feature points which results the feature matching time is too long, as a result, the algorithm reduce the matching efficiency. The paper put forward a kind of license plate recognition based on PCA-SIFT algorithm, which make the process of the license plate recognition more quickly by reducing the dimension of feature points.The system includes three parts:license plate image pretreatment, license plate location and character recognition. The license plate image pretreatment part contains gray-scale transformation, filtering, sharpening, normalization and modifying the parameters of the traditional PCA-SIFT algorithm. Experiments show that modifying the parameters of the PCA-SIFT algorithm improves the accuracy of plate positioning. In the license plate location part, first extract the PCA-SIFT features of the standard positioning plate and save them, second, extract the PCA-SIFT features of license plate to be identified and save them, next, match these two features, and last, segment the license plate according to the results of the feature point matching. In the character recognition part, first extract the PCA-SIFT features of the standard template characters and save them, then extract the PCA-SIFT features of the positioned license plate, at last, characters are identified by matching the feature vectors. This paper applys MATLAB7.1in simulating the algorithm, compared with the traditional algorithm, a number of experiments show the algorithm has a fast calculation speed and a higher recognition efficiency.