Authors: R. Veerasundari, S. Umamaheswari
Abstract: Fusion methods can be used to produce high resolution multispectral images from a high resolution panchromatic (PAN) image and a low-resolution multispectral image (MS) for the spatial extent. The fused image will have structural details of the higher spatial resolution panchromatic images as well as rich spectral information from the multispectral images. A small portion of Chennai city is considered as test data for the analysis. Before fusion, Mean adjustment algorithm of Adaptive Median Filter (AMF) and Hybrid Enhancer (combination of AMF and Contrast Limited Adaptive Histogram Equalization (CLAHE)) are used in the preprocessing. Here, conventional Principal Component image fusion method will be compared with newly modified Curvelet transform image fusion method. Principal Component fusion technique will improve the spatial resolution but it may produce spectral degradation in the output image. To overcome the spectral degradation, Curvelet transform fusion method can be used. Curvelet transform uses a curve which represents edges and extraction of the detailed information from the image. In Curvelet Transform, individual acquired low-frequency approximate components of PAN image and high-frequency detail components of MS image are used. Peak Signal to Noise Ratio (PSNR) and Root Mean Square Error (RMSE) are measured to evaluate the image fusion accuracy.
Keywords: Image De-noising; Image Enhancement; Image Registration; Image Fusion; PCA; Curvelet