Investigation of Feature Extraction Methods for Image Retrieval Application

Abstract

Content-based image retrieval is a technique used for retrieval of desired images via their colour, texture, and shape features. Features play a major role in an image. The major challenge in image retrieval lies in extracting the optimal features from an image. Feature extraction is a process of selecting optimal low level feature subsets. It transforms the input image into a set of features that describes the image with sufficient accuracy. In this paper, three specialized features i.e. colour moments, Region properties and Grey Level Co-Occurrence Matrix (GLCM) are extracted. This Image retrieval system using the hybrid features are tested using Corel image datasets consisting of 1000 images from 10 semantic categories. The efficiency of the system is evaluated in terms of precision, recall and error rate. From the experimental results, we can conclude that these hybrid features have improved the precision of the retrieval system when compared with other state-of-the-art methods.