GLCM Feature Extraction for Visual Speech Recognition
Ajit S. Ghodke1, Yuting Zhang1, Ritesh A. Magare2
1International Education College, Neusoft Institute Guangdong, Foshan, 528225, China
2Department of Computer Science and Information Technology, Dr. Babasaheb Ambedkar Marathwada University, Ma-harashtra State, 431004, India
Abstract: Visual data based speech recognition is a field with a great potential to help solve difficult issues like audio corruption by noise. This paper addresses an application of various GLCM features to extract re-gion based features for motion estimation of images. The features like contrast, energy, homogeneity, mean, standard deviation, entropy, variance, smoothness and IDM have high discrimination accuracy and requires less computation time and hence efficiently used for pattern recognition like speech recognition application. In this paper we used the visual dataset which is prepared by Samsung galaxy nxt mobile phone.
Keywords: GLCM (Gray Level Co-Occurrence Matrix); IDM(Inverse Different Moment); Feature extraction; Visual speech recognition