Machine recognition of Brahmi script

Abstract

Computer recognition of machine-printed characters of the Brahmi alphabet is described. Each character is converted manually into a rectangular binary array in which a zero represents a blank, and a one, a nonblank. Features in the form of strings are extracted by row-wise and columnwise scanning of the array, depending upon the number and nature of the runs of ones in the rows and columns. Feature strings of the characters of the Brahmi alphabet form a dictionary. A given text is presented character by character, and features are extracted in the form of strings and compared with the strings in the dictionary. When there is perfect agreement, the letters are recognized and printed in Roman, following a special method of transliteration.