PRODUCING HIGH RESOLUTION SPECTRAL BANDS FROM LOW RESOLUTION MULTI-BANDS IMAGES, USING PRINCIPAL COMPONENT ANALYSIS “PCA” TECHNIQUE

Abstract

Image data fusion is the process of setting together information gathered by different heterogeneous sensors, mounted on different platforms. This research presents an effective multi-resolution image data fusion methodology, which is based on utilizing the Principal Component Analysis “PCA”. The first principal component “PCA1” involves much of the variability in the spectral data; while the reminder PCAs contain the remaining variability in a descend order. The low resolution multispectral bands are, firstly, resized (i.e. enlarged) into the high resolution “panchromatic” image size, then transformed into several PCAs. As first step the panchromatic image is normalized to have the same number of gray levels as the PCA1, then replacing the PCA1 of the low- resolution-multispectral image in the PCA transformed domain. The high-resolution-multispectral images are produced by inversely transform the modified PCA’s file.