Research Of Character Recognition

Digital image processing and pattern recognition are hot topics in computer application field, and most of the researchers are always pursuing the ways to make computers do the work instead of human, the characters image recognition is one of those work, and it is widely used in handwriting recognition, intelligent traffic and so on, it is still important meaningful to do research about character recognition. This article made research of character recognition, the mainly job including:First, pre-processing step of source image is researched, including turn the source RGB image to gray image, image filtering, binarization of gray image, separate characters into every single character, thinning the character image into one pixel wide, every step is realized by MATLAB, and analyze the influence and result of each step according to experiment;Second, researched the character recognition methods, the temple match is easy to realize, in this article, character recognition is realized by two kinds of temple match, one compare the character and the temple directly, one compare the cross times feature of character, and then combine these two method to improve the efficiency;At last, combined these steps, and realized the verification code recognition and license plate recognition, then according to these two applications of character recognition, analyze the influence of each step to the final recognition result, and summarize how to choose the recognition method by the feature of the source image.