Introducing an efficient method for feature extraction in image retrieval systems

Image retrieval systems are of particular importance because they can be effectively used in image management in large databases as well as in search engines. By enabling fast and accurate image searches based on relevant features, these systems help users quickly access the desired information and significantly increase the efficiency of search processes. However, current systems usually do not cover the comprehensive features necessary to describe different types of images. In this research, we have tried to improve the performance of image retrieval systems in this aspect by using a wider set of features. Our proposed method generally consists of several steps. In the first step, we perform the feature description process, in which four feature categories are used to describe image characteristics. These categories include texture, color, and various patterns related to these two features. Then, a feature selection algorithm based on Black Hole Optimization (BHO) is used to extract an optimal subset of these features. This step helps us to improve the retrieval accuracy in addition to increasing the processing speed. Finally, a Fuzzy C-Means (FCM) is used to retrieve images based on the selected feature set. The performance of the proposed method was evaluated based on two datasets and achieved significant precisions of 0.97, 0.95, and 0.94 in the Corel1000, Brodatz, and ALOI databases.