Projection profile based ligature segmentation of Nastaleeq Urdu OCR
Projection profile based ligature segmentation of Nastaleeq Urdu OCR
复制标题
基于投影轮廓的 Nastaleeq 乌尔都语 OCR 连字分割
DOI：
10.1109/iscbi.2016.7743278
复制
发表时间：
2016
期刊：
影响因子：
--
通讯作者：
Ajay Koul
中科院分区：
文献类型：
--
作者：
Aejaz Farooq Ganai;Ajay Koul
乌尔都语Nastaleeq是一种高度草书、上下文敏感的语言，从右上角到左下角对角书写。这使得难以将部分单词或完整单词分割成字符。此外，由于字符的堆叠，难以执行字符级的分割。一些研究人员已经进行了分割，并取得了很大的成功，但仍有一些类别的乌尔都语字母尚未被识别。此外，分割的准确性仍然是不够的，必须提高发展光学字符识别。我们提出了一种新的投影轮廓分割方法的乌尔都语Nastaleeq连字。在细分过程中遇到的各种挑战进行了详细讨论。
Urdu Nastaleeq is a highly cursive, context sensitive language, written diagonally from top right to bottom left. This makes it difficult to segment the partial word or a complete word into characters. Further due to stacking of characters, the segmentation at the character level is hard to perform. Some researchers have performed the segmentation and have succeeded to a good extent, but still some classes of Urdu alphabets have not been recognized yet. Moreover accuracy of segmentation is still less and has to be improved to develop Optical Character Recognition. We present a Novel approach of Projection Profile methodology for segmentation of Urdu Nastaleeq Ligature. The various challenges encountered during segmentation have been discussed in detail.