An Improved IHS Image Fusion Algorithm using Medoid Intensity Match and Bilateral Filter

Image fusion is a productive way to combine multi-sensor images and extract maximum information from enhance remote sensing data. The paper describes a novel methodology to improve IHS image fusion algorithm by medoid intensity computation from multispectral images and suppress noise of panchromatic image using bilateral filter. Medoid provides an optimal representation of multispectral images and act as an intensity component in IHS color space. Bilateral filter preserves the edge information of high resolution data while reduce the noise in homogenous regions. Improved IHS image fusion algorithm by amalgamation of medoid intensity and bilateral filter found to be an effective technique to generate fused images. The technique developed is evaluated with Cartosat-1 panchromatic and Resourcesat-2 multispectral closest acquisition images. The datasets are geometrically transformed and co-registered to make it eligible for image fusion exercises. The visual assessment confirms improved IHS better retain the spatial details while preserving the spectral characteristics. Improved IHS is compared with state-of-the-art component substitution image fusion techniques and found to be perform better in almost all the standard metrics.