The Study On The Character Image Recognition System Based On Image Processing

Many files exists with the form of the paper, for example bank papers, tax administration statements, the standardize machine reading cards in the examination,census forms, allots, order forms etc. But the paper files are inconvenience in conserving,indexing, Star., and amending. This is a nuisance that putting these information into computer. During a long time, people input the information according to the hands, but this way is taking more times and using a great effort. At the real time instance, it is very important to process the files automatically, fleetly rule and line. So, it is very important to put the files into computer automatically.In this paper the problem of computer recognition to the character image is studied. According to the style of writings, the characters are divided into two forms: printing characters and handwriting characters. Based on every characteristic of the two styles,system realizes the function of image pre-processing, characteristic extraction and character recognition. So the whole system is divided into four parts, image collection module, image pre-processing module, characteristic extraction module and character recognition module.According to the individual needed of printing and handwriting character processing,two different strategies are adopted by the system. For the printing characters, first, the system completes the image processing and then extracts the combined characteristic of the net characteristic and the intersection characteristic. Basing on these a standard characteristic database of characters is established. And for the handwriting characters, we adopted the method of getting the outline characteristic of the characters.The recognition arithmetic of printing character adopts template matching, which compares the character with the standard characteristic database and recognize the character as the nearest Euclid distance standard character in the standard characteristic database. And for the handwriting character, the system adopts BP neural network based on Genetic Algorithm (GA). The purpose of GA is to make sure the initialization power value and increase the preciseness rate of the recognition.The software development adopts the NI Company's Virtual Instrument Lab view and IMAQ Vision tool. The software realizes the function of image collection, image pre-process, character extraction and character recognition. The system successfully realizes the BP neural network based on GA in Lab view. It has the merit of short period of impoldering, convenient programming, high repeated using rate, easily operation, friendly interface, high reliability and so on. The experiments show that the recognition accuracy of printing numbers is 98.96%, the recognition accuracy printing letters is 98.35%; and the recognition accuracy of handwriting numbers is 94.49%, the handwriting letter is 89.13%.