Main Article Content
This paper shows a new way to use a combination of filter feature selection algorithms for image classification. In pattern recognition, machine learning, and computer vision, feature selection is one of the most important problems. The main goal of feature selection is to classify images, improve how well they can be classified, and make the whole process easier to understand. The only method that is guaranteed to find the best subsets is the exhaustive search method, but it takes a lot of time to run. A new adaptive and hybrid approach to selecting features is proposed. This approach combines and uses different methods to make a more general solution. Several state-of-the-art feature selection methods are described in detail with examples of how they can be used, and a thorough evaluation is done to compare their performance to that of the proposed approach. The results show that the individual methods for selecting features perform very differently on the test cases, but the combined algorithm always gives a much better answer.