This work proposes a cluster based hierarchical technique for recognition of handwritten Bengali character. Due to the consecutive appearance of Bengali text segmentation of Bengali character is not very easy. | ISSN:2249-5789 Satarupa Bagchi Biswas et al , International Journal of Computer Science & Communication Networks,Vol 2(5), 590-595 A Cluster Based Hierarchical approach for the recognition of Bengali Handwritten Character Satarupa Bagchi Biswas1 , Smritikona Barai2, Sandipan Dutta3 1 .Asst. Professor, Dept. of IT, Heritage Institute of Technology, Kolkata. 2 .Asst. Professor, Dept. of IT, Heritage Institute of Technology, Kolkata. 3. .Asst. Professor, Dept. of IT, Heritage Institute of Technology, Kolkata Abstract This work proposes a cluster based hierarchical technique for recognition of handwritten Bengali character. Due to the consecutive appearance of Bengali text segmentation of Bengali character is not very easy. For successful recognition of handwritten Bengali character proper segmentation is an important criterion. In this respect the present approach trying to provide a complete solution for Bengali handwritten character recognition. Here mainly we will concentrate on single Bengali character recognition. Keywords: Bengali Handwritten character recognition, Clustering, OCR, Segmentation. 1. Introduction Bengali is the second most used language in India. Moreover certain languages like Manipuri, Ahamia have the similar script like Bengali and also Bengali is the official language of Bangladesh. Successful recognition helps to office automation and saves huge amount of time and effort. Though there are lots of commercially available system is there but yet they can be further extended to handwritten text. Thus handwritten character recognition research for Bengali script has a lot of significance. Research works on optical character recognition (OCR) for printed Indian scripts including Bengali [1] are found in the literature. A survey of Indian script character recognition research is also available in [2]. However there are not very significant research works done so far. Unfortunately the technology used for printed character cannot be extended
