Script identification based on wavelet energy histogram moment features

In this paper, we propose a novel texture feature extraction method which compute the weighted moment of the wavelet energy histogram of the decomposed images. A total of 15 features are extracted for script classification. The author choose six languages (Arabia, Chinese, English, Hindi, Thailand and Korean) to demonstrate the potential of the technology. Experimental results show that the propose method has less number of features and is more effective in script identification than the traditional wavelet energy and ratio method and histogram signatures algorithm. © 2010 IEEE.