A line of a bilingual document page may contain text words in regional languageand numerals in English. For Optical Character Recognition (OCR) of such adocument page, it is necessary to identify different script forms before running anindividual OCR system. In this paper, we have identified a tool of morphologicalopening by reconstruction of an image in different directions and regionaldescriptors for script identification at word level, based on the observation thatevery text has a distinct visual appearance. The proposed system is developedfor three Indian major bilingual documents, Kannada, Telugu and Devnagaricontaining English numerals. The nearest neighbour and k-nearest neighbouralgorithms are applied to classify new word images. The proposed algorithm istested on 2625 words with various font styles and sizes. The results obtained arequite encouraging
展开▼
