Improve recognition-based segmentation for printed persian character recognition

Abstract

The most scientific papers dealing with handwriting recognition systems make statements relating to recognition performance based on a forced-recognition rate. This rate describes the ratio between the number of the correct recognized samples and the number of all possible samples. This paper proposes a new technique to improve the previous research work in order to have a better result. As part of the preprocessing phase the image file is checked for skewing. If the image is skewed, it is corrected by a simple rotation technique in the appropriate direction. Character fragmentation process is done based on the existing technique. The improvement of feature extraction process is made by detecting details of glyph and used as an input for the classification process of ANN (Artificial Neural Network) to develop a full OCR system. Result indicated that some versions, even upon first using them, were at least as good as last methods of recognition.