Recognition of ancient characters in stone inscription

In our project, we simulate a technique for recognizing ancient characters in stone inscription of various centuries. Many medical mysteries and valuable historical secrets are hidden in stone inscription which only few people can understand. Though there are number of character recognition, there is no efficient system for language which has a large character set and huge amount of variation. In particular, there is no system that addresses either the morphological characteristics of stone inscriptions or accounts for ancient characters. We propose a 3-stage approach to increase the recognition efficiency of characters. The image of the sculpture is captured and converted to required dimensions. Segmentation of the character is done using Chan-vese algorithm which uses active contours for separating the characters from background. Chan-vese algorithm has resistance to the morphological characteristics of stone inscriptions as it uses active contours. The noise in the image is removed using the area property. Zoning of characters is done by which each character is divided into six regions each viz., three horizontal and three vertical regions. Feature extraction is carried out by extracting the geometric features of the character contour. These features are based on the basic line types that form the character skeleton. 54 features (nine for each region) representing the number and normalized length of line segments of is extracted. The feature vectors so generated are applied to neural network (which is trained in advance with the features extracted from the characters of training database) for pattern recognition. Post processing of the pattern obtained is done to identify the exact character.