Texture Classification Using Wavelet Packet Decomposition Based on SGS & MISS Algorithm

This paper describes effective texture characterization techniques using wavelet packet decompositions. The proposed work presents a statistical view of the image classification problem by combining the tasks, namely feature extraction (FE) and similarity measurement (SM), into a joint modeling and classification scheme. The proposed method uses the wavelet packet energy as a feature for texture classification for a compact feature representation and better classification accuracy. It shows that incorporating the sub band grouping and selection method and Fuzzy C Means algorithm improves the feature selection and classification accuracy. Experimental results on a database of texture images indicate that the new method significantly improves Classification rates compared with traditional approaches, while it retains comparable levels of computational complexity.