Design And Implementation Of ID Card Recognition System Based On Image Processing And Pattern Recognition

Optical camera image processing and pattern recognition theory based on optical character recognition(OCR) has been widely used in various sectors of social life. Optical character recognition has the advantages of fast speed, high efficiency, low cost, less errors in the text input. Certificate recognition developed based on optical character recognition as an emerging technology in every aspect of social life has been applied.The main research contents involved in identity information recognition software based on optical image was carefully analyzed based on the research in the domestic and foreign research present situation; the identity information recognition has been designed software based on MATLAB platform.Image preprocessing is the first step of ID identification software based on optical image. For the optical image of the ID, the average color offset was used to correct the high light areas of an image. On the basis of T.Youg theory, the ID card image was graying by using the weighted average. Binaryzation image was gotten by Ostu-Berensen method combined by the global binaryzation and local binaryzation. Finally, layout analysis of the binaryzation ID card image was done by using the method of projection and column method, character block images were gotten to be used for character recognition.For the problem of different image size and location after pretreatment, the center normalization methods of gravity position normalization, frame normalization and the size normalization method using normalized variance of distribution were analyzed. The 7 x 7 x 4 Chinese characters were formed by fuzzy stroke feature within global elastic mesh using a combination of statistical information and structural information for character recognition.After extraction of single character image through character feature vector normalization processing, character feature extraction of image preprocessing, the feature vectors needed to get into Chinese characters classifier to complete identity information. Image classifier includes the Euclidean distance classifier, the neural network classifier, support vector machine classifier and genetic algorithm classifier.With the detailed analysis of the image preprocessing, feature extraction method and the method of character classifier design, the identification software based on optical image on the platform of MATLAB was designed. The key function prototype and the simple process of function of the ID card identification system were discussed in detail. Functions include image preprocessing, images binaryzation, image character partition, character feature extraction and character recognition in detail design. Finally, the recognition result on the ID card recognition software based on the optical image of designed in this project showed that the recognition probability of the system in the identity information of gender, ethnic, birthday, ID number is higher, while in the two aspects of the name and address of the recognition ability has yet to be improved.Through this topic research on the key technology of identity information recognition software based on optical image, the information identification software system was designed on MATLAB platform. The software has realized the high identification probability of gender, ethnic identity cards, birthday, ID number, and the probability of recognition including name and address relatively low. Identity information recognition software optical image to be applied widely needs requirements of a cost control and the software running time. Identity information recognition method on low resolution still needs further research. Identity information recognition software algorithms still need to be further optimized, to achieve rapid extraction ID information purposes.