Script line separation from Indian multi-script documents

In a multi-lingual country like India, a document page may contain more than one script form. Under the three-language formula, the document may be printed in English, Devnagari and one of the other Indian official languages. For Optical Character Recognition (OCR) of such a document page, it is necessary to separate these three script forms before feeding them to the OCRs of individual scripts. In this paper an automatic technique of separating the text lines is presented for almost all triplet of script forms. To do so, the triplets are grouped into five classes according to their characteristics, and shape based features have been employed to separate them without any expensive OCR-like algorithms. The proposed approaches are tested-on many documents and the experimental results are presented. At present, the system has an overall accuracy of about 98.5%.