An Offline Fuzzy Rule-Based Classification System for Persian/Arabic Handwritten Word Recognition

An algorithm for recognition of offline Persian handwritten word images is presented based on Fuzzy Rules. We described three methods for feature extraction: the first method is based on the shape of the boundary of the word image and the second one is based on the position of non-subwords in the word image, and third one extracts holes. The proposed feature extraction methods are scale and rotation invariant and also good to withstand noise and incomplete information. Experiments carried out on 9,326 lexicon size showed promising performance results of 81.73%.

کلیدواژه ها:

Word Recognition ، Fuzzy Rule-based Classification System (FRBCS) ، Subword Extraction ، Hole Extraction ، Persian Handwritten Recognition