Thai Font Type Recognition Using SIFT

Abstract

This paper presents a Thai font type recognition on Thai document by using Scale-invariant feature transform (SIFT). The features are extracted by Scale-invariant feature transform (SIFT) that is widely used in image processing. Sift is an algorithm for detecting local features in order to find similar objects. Our system contains ten fonts and ten text images in each font. We use ten text images each font total one hundred images for our experiment. Our results show accuracy for 97.37% for ten Thai fonts.