English Character Feature Extraction
Jin Wei-jie · Jisuanji fangzhen · 2007
English character recognition is an important branch of pattern recognition,and it is worth studying.This system consists of document segmentation,word segmentation,character recognition and post-processing.This system implements this whole process from image scanning to getting the result of the documents recognition.Feature extraction is a primary content of the system,and its research object is twenty six capital letters and twenty six lowercases.Character recognition is consisted of image preprocessing,feature extraction,model building,classifier design and post-processing.In the article,several features widely used in OCR,such as gridding feature,through feature and distance classifiers are explained and compared,then many experiments are made.The results of the test experiment show that this system is with well performance.