Research On Railway Freight Car Licence Recognition Algorithm

Railway freight car licence recognition is one of important applications in image processing and recognition.The plate number is often become faint and incomplete because it was worn or defaced during used in the field, and the plate number is often adhesion and fracture, so it is less common the identification system based on machine vision. In this paper, the algorithms of railway freight car licence recognition are researched. The recognition rate is improved, and it meets the requirements of practical application. The main research contents are as follows:1. In the process of license plate image preprocessing, the algorithm of plate location, the algorithm of image denoising, the algorithm of image binarization is studied. In the image binarization processing, the image was binarized using the combining algorithm of Otsu algorithm and histogram algorithm.2. For the phenomenon of adhesion and fracture in the railway freight licence, the segmentation algorithm based on projection, the segmentation algorithm of shortest path, the segmentation algorithm based on identification were studied in the processing of touching character segmentation. The algorithm of segmentation search tree and the arc characteristics segmentation algorithm were studied in the fracture character segmentation. Through the analysis of the railway licence area structure, the touching characters were segmented using the combination algorithm of shortest path algorithm with projection segmentation algorithm, and the fracture characters were segmented using the arc characteristic and the width of the plate number. The segmentation of railway freight licece plate is achieved.3. The common feature extraction methods are described. The feature extraction method based on combination of Hu invariant moments and digital figures features is presented. 7 Hu invariant moment guarantee the rotation, translation invariance and scale invariance of the characters. The concave feature, the center line feature, the cicular feature enhance the difference between the different characters.The common recognition algorithms of digital are studied. The template matching method and BP neural network method for railway number identification was test, and finally the railway freight car licence was indentified using the BP neural network.4. The recognition system of railway freight car licence was designed. In VC6.0 environment, some of OpenCV library functions were called. The sysem includes image processing module, feature extraction module, number recognition module.