Studying Vehicle Sales Invoice Recognition Algorithm

With the rapid development of computer science, pattern recognition, digital image processing as an important branch of the field, through the development of computer software and hardware changes in the east, in recent years ushered in the development and promotion of high-speed in urban traffic management, postal class mail automatically sorting, as well as tax, finance, insurance, banking instruments widely used in various industries and graphics recognition, there are many successful applications, so that faster and better development of these industries.In most government agencies or service industries, along with the database is a huge paper-based information repository, and paper-based information repository for electronic management, administrative authorities gradually moved toward a paperless, it is the most popular needs and applications .In fact, E-paper files can bring us great benefits and convenience, but the process of electronic is painful, because the data collection process is quite cumbersome. Using pattern recognition technology, you can put the entire process electronically, including images of the collection, classification, treatment, identification, preservation, intelligent, automated, off, or only a small amount of human intervention, so that such systems are pushed smooth and applied continuous. Extracting key information from the image, is the process of image recognition.In order to solve the file electronic tax collection computer tasks faced by a large amount of inaccurate data entry problems of the unity of the motor vehicle sales invoice recognition algorithm made some research, and preliminary identification process to solve various problems. Subjects in research is a unified vehicle sales invoice recognition algorithm by analyzing the layout and structure of the invoice, the invoice image using Radon transform to detect lines in the table and tilt the image accordingly to solve the problem; and continue projection, area and other methods to achieve the invoice number positioning, segmentation areas of interest; final design of the three kinds of simple feature extraction method for the identification number to complete the task.