Bangla Character Recognition by Euclidean Distance Between Center of Gravity and Endpoints

Abstract

Character recognition is one of the classic examples of pattern recognition and image processing. Many works has been carried out by researchers in this field. Bangla character recognition becomes an interesting field of research in last few decades. However, researches on Bangla character recognition still in progress. This paper contributes an approach toward character recognition by calculating Euclidean distance between center of gravity (CoG) of a character image and its endpoints. The CoG is calculated based on a research work by Onodera (Singapore ICCS/ISITA, 2:548–552, 1992, [1]). Endpoints of a character are generated from skeletonized image followed by removal of isolated pixels and CoG. Parameters like number of endpoints, mean of normalized Euclidian distance and mean CoG are used for recognition. The above-mentioned method is applied on handwritten characters of different sizes and width with satisfied recognition accuracy.