Recognition of Machine Printed Broken Oriya Characters Using Sift Features
Sandeepika Sharma, Sneha Choudhray, Bhupendra Kumar · 2015
OCR for Indian scripts has been an active area of research for years. The fine printed documents can be recognized easily but the technology underperforms in case of degraded document images. In this paper a gradient based feature extraction technique is presented to recognize the broken printed Oriya characters. Scale Invariant Feature Transform (SIFT) key point descriptors are used as the feature set to represent each character sample at the feature space and a Brute-Force Matcher interface is used as a classifier to classify the character images. On the basis of the SIFT feature set the recognition rate was found to be 82.7%