Fusion multispectral images by fuzzy-ARTMAP

Abstract

Remote sensing satellites produce images in two modes: panchromatic with very high spatial resolution, and multi-spectral mode with a spatial resolution lower. The purpose of this paper is to develop a method which combines the two modes. We built a new image with the spectral properties of the low-resolution image and the spatial detail of the high resolution image. The images used are derived from IKONOS and QuickBird satellites with a ratio four (4) for the resolution. The basic idea of this work is to develop and test the supervised fuzzy ARTMAP of adaptive resonance theory [1, 2]. In this context two network fusion methods and IHS fusion process combined with a "histogram matching" have been developed. The quality of fusion was done qualitatively and quantitatively by taking into account the two aspects: the quality of preservation of details and quality of color conservation. Although improvements are still needed, the approaches developed and put into practice the evaluations have yielded promising results.