Improvement of Bat Algorithm Classification Accuracy Using Image Fusion Techniques

This paper investigates the pansharpening influence on satellite images-classification using Bat algorithm (BA). To this end, experiments are proceed using two fusion techniques: Brovey Transform and Intensity-Hue-Saturation transform, in order to merge the characteristics of images of the same area. Considering the classification as an optimization problem, BA can be applied on a fully-featured image. For this research, recent Landsat 8 panchromatic and multispectral images taken over the city of Oran (Algeria) are used to show the performance of BA and the benefit of using fusion techniques to improve classification. This paper shows improvement in the results when a fusion step is applied. Additionally, BA performance is compared against K- Means and Particle Swarm Optimization. From the obtained results, it can be concluded that BA can be successfully applied to solve unsupervised classification problems.