Adaptive Hybrid Remote Sensing Image Fusion of Panchromatic and Multispectral Images
Image fusion is one of the most fundamental methods used in advanced image processing for growing demand to produce an accurate, clear and realistic images. This technique often performed to preserve the spatial information and also to avoid spectral distortion. In this paper, the proposed system comprises of two different approaches for adaptive pan-sharpening algorithm for image fusion. In MRA pan-sharpening algorithm, low frequency components are removed using gaussian filter and the obtained high frequency components are injected into the MS image with suitable gain factor. In adaptive hybrid pan-sharpening, the primary high frequency components evaluated by taking the difference between histogram of PAN image and intensity component of MS image. Secondary high frequencies are obtained by taking Laplacian between histogram of PAN image and intensity component of MS image. The high frequency obtained is injected into the MS image with suitable statistical similarity pixel ratio to avoid spectral distortion. The datasets used in this proposed system are IKONOS and Quickbird satellite images. The fused images are compared with existing system which provides a better quality of images under perceptual and qualitative assessment.
Keywords: Image fusion; Spectral distortion; Gaussian function; Smoothing factor; Histogram; Gain factor