الفهرس | Only 14 pages are availabe for public view |
Abstract The objective of tlus thesis is to discuss the problem of Arabic online handwriting recognition and finding techniques to be applied in a system for recognizing Arabic characters and words. At first, the research was focused on the recognition of handwritten isolated Arabic characters and Hindi numerals which are used in our daily life as Arabic numerals. Later on, a prelimimuy study is ca1Tied on the recognition of handwritten Arabic words (which is cursive by its nature). This study results on developing a word segmentation module to be used in word recogmtwn system. The study also discussed the needed modifications on the otl1er modules of tl1e word recognition system to be compatible with the output from the developed word segmentation module. Most of the described techttiques are implemented in the developed programs within tlus tl1esis. The major stages of the introduced handwriting recognition system are as follows:The classification (Segment or character) is done with the aid of fuzzy rules. The fuzzy rules simplify the design and provide the ability to extend the rule base easily, which provides the system the ability to extend the set of defined characters to be recognized. Many ambiguities that exists in handwritten characters can be reduced by the techniques discussed in this thesis. These techniques are flexible enough to overcome the variations of handwriting modes. The flexibility of these teclmiques permits the recognition system to be able to recognize characters from a new writers. |