An Approach for Segmentation of Handwritten Documents Using ICA Algorithm

There are numerous governmental, social, business, and educational associations that manage large number of handwritten documents. As time proceeded, the content in the handwritten document gets blurred and that cannot be readable. In order to make that text visible and accessible, Document Image Analysis (DIA) is used.  The previous work deals with the pair wise similarities between word-separators as well as unary properties in the document. But there is a failure case, when both the inter and intra word gap is same. Then structured SVM fails to work. The proposed system uses Independent Component Analysis (ICA) algorithm, which transforms multivariate data so as to make its essential structure more visible or more accessible, thus facilitating the analysis of the data. Here the Tamil handwritten document is taken into consideration since the local language is Tamil.