An Expert System Based on Discrete Wavelet Transform - ANFIS For Acquisition and Recognition of Invariant Features from Texture Images

Abstract

Texture is low-level image features. Today, texture analysis is applied to different fields such as ultrasound images recognition, document classification, classification of radar imagery, texture-based image retrieval. Texture analysis and classification is an important issue. Comparison of texture analysis and classification was characterized by the lack of adequate methods. Wavelet decomposition method is superior to other conventional methods. In this study, Discrete Wavelet Transform (DWT) Adaptive Network Based Fuzzy Inference System (ANFIS) were used for acqusition and recognition invariant properties of tissue-type image. 22 texture images were taken from Brodatz database for the analysis of tissue-type image. 50 randomly 64 x 64 regions are selected from each of 22 images. Testing Succes with the average rate of 91.27% was obtained by using DWT-ANFIS method.