Ligature Segmentation for Urdu OCR

Urdu script uses superset of Arabic alphabet, but uses Nastaliq writing style. Nastaliq script is highly cursive, context sensitive and is written diagonally from top right to bottom left with stacking of characters, which makes it very hard to process for OCR. In addition, line and word segmentation are non-trivial tasks as we have frequently merging lines and vertically overlapping words and ligatures. Due to the challanges in character segmentation most of the researchers have taken the next higher unit, ligature, as recognition unit. A ligature is a connected component of one or more characters including diacritic marks and usually an Urdu word is composed of 1 to 8 ligatures. In this paper, we present a methodology for segmenting the Urdu text into ligatures. A hybrid approach, which uses top down technique for line segmentation and bottom up design for segmenting the line into ligatures, has been employed. The various challenges encountered during ligature segmentation such as horizontally overlapping and broken lines, merged ligatures and diacritic association have been discussed in detail.