The Study And Development Of The International Mail Translation System

The translation of the mail address is correct or not depends on the recognition. Reducing the refuse recognition rate and error recognition rate of the mail address can effectively improve the translation rate and translation accuracy rate of the International Mail Translation System. With the development of science and technology, more and more people pay attention to Optical Character Recognition Technology. It integrates digital image processing, computer graphics, intelligence and other aspects of knowledge, improves postal business labor productivity.This paper is based on the International Mail Translation System, studying the character recognition of images for International mail. It integrates the envelope characteristic and develops a set of effective image preprocessing algorithm. The binary image segmentation is realized through achieving optimal threshold by Otsu algorithm. The inclination adjustment of the image is realized by least square method. The character segmentation and normalized adjustment is completed based on the outline of the character. According to the characteristics of the print English letters and Arabic numerals, the grid characteristics and statistical characteristics are extracted as recognition characteristics to improve the recognition rate of the system through the combination of eigenvector.According to the character features of the address, the number of the input layers, output layers and Hidden neurons are confirmed, the parameter design of BP neural network is discussed in detail. Finally, the processing of training and recognition of BP neural network analyzed significantly includes the selection of the training samples and the requirement of the training cease.This paper develops the software test platform of the system according to Visual C++ 6.0 and OpenCV library and do the recognition test through lots of English mails. It concludes that the recognition rate is more than 98%, the refuse recognition rate is 0, and the desired objectives of the system are reached.