Image retrieval based on Texture Feature Extraction using GLCM & Wavelets

Abstract

Retrieving images in an ever increasing and large image datasets has become an exigent task forcomputer vision and pattern recognition applications. In this article, a statistical method thatconsiders the spatial relationship of pixels i.e., the gray-level co-occurrence matrix (GLCM), alsoknown as the gray-level spatial dependence matrix is used to identify texture features. PCA isapplied to reduce the dimensionality of data and build train feature matrix. Five different distancemeasure techniques are used for classification purpose. Proposed method is demonstrated on a verylarge yet challenging Caltech-101 dataset and observed the impact of GLCM texture features interms of improved classification accuracy in comparison with some of the popular techniques foundin literature.