A novel image retrieval model based on the most relevant features

This paper presents a proposed model for content-based image retrieval (CBIR) which depends only on extracting the most relevant features according to a feature selection technique. The suggested feature selection technique aims at selecting the optimal features that not only maximize the detection rate but also simplify the computation of the image retrieval process. The proposed model is divided into three main techniques, the first one is concerned with the features extraction from images database, the second is performing feature discrimination, and the third is concerned with the feature selection from the original ones. As for the first technique, the 3D color histogram and the Gabor filter algorithm are used to extract the color and texture features respectively. While the second technique depends on a genetic algorithm (GA) for replacing numerical features with nominal features that represent intervals of numerical domains with discrete values. The GA is utilized in this technique to obtain the optimal boundaries of these intervals, and consequently to reduce the complexity in feature space. In the third technique, the feature selection performs two successive functions which are called preliminary and deeply reduction for extracting the most relevant features from the original features set. Indeed, the main contribution of the proposed model is providing a precise image retrieval in a short time.