Efficient Image Retrieval Based on Support Vector Machine and Genetic Algorithm Using Color, Texture and Shape Features

Abstract

Content based image retrieval (CBIR) systems can find similar images to a query image in a large image database. This technique is based on the visual features of the image. In this work we propose a CBIR system based on the three descriptors of the image which are the color, texture and shape features. This study extracts robust features from all dataset images and the query image with the same manner. The image descriptors are extracted from the color histogram, gray level co-occurrence matrix and the Hu moments. Then, a classification technique based on the support vector machine is applied to the features database to create four image classes in the purpose of reducing the query time and limiting the search interval. Meanwhile, the image retrieval is performed based on an efficient meta-heuristic algorithm which is the genetic algorithm. The precision and recall measurements are computed based on the obtained results to validate the efficiency of our CBIR system.