|
چکیده
|
Many methods have been suggested for recognizing handwriting in different writing systems in which various strategies and techniques are used such as statistical methods, structural methods, neural networks and etc. Applying each one of these methods require some preprocessing in order to prepare the primary data for the main processing. The present article, benefits from previous researches in Persian OCR, Arabic, English handwriting and etc. Although researchers in Iran and different countries have done great efforts to recognize Arabic handwriting, yet because of the differences and unique structure of Kurdish handwriting, it was necessary to discuss it specifically in a different article. In this article, we discuss some methods for extracting the features of the letters to categorize them. These could be used in recognizing Kurdish handwritten letters. First, we briefly point out the differences of Kurdish letters with Persian and Arabic letters; next, we discuss the applied features and structures. Then, we suggest some methods to extract these features; and finally, we categorize the letters blending these features. This work could be a solid ground for further research in recognizing Kurdish handwriting.
|