This paper proposes segmentation based optical word recognition system, where the segmentation is a geometric technique for extracting features. Fostering of feature vector is done from the raw form of character skeleton, which is used to determine basic line types. Features of the given scanned input image are then compared with feature vectors so generated from the training sets. This technique also concentrates on positional features. With the help of these feature vector, pattern recognition engine can be tutelage based on Neural Networks.
Volume 12 | Issue 2
Pages: 727-733
DOI: 10.5373/JARDCS/V12I2/S20201089