STUDY OF IMAGE FUSION TECHNIQUES FOR MULTI-RESOLUTION DATA SET

Abstract

This study deals with the review of image fusion techniques in which some are spatial domain based and some are frequency domain based. Out of them the spatial domain based image fl.ision techniques such as Brovey Transform (BT). Flue Saturation Intensity (IISI) and Principal Component Analysis (PCA) are applied on five data sets using MATLAB software as the tool. Among them four data sets are of single sensor and one data set is of multisensors. l'rue Colour Composite (TCC) and False Colour Composite (FCC) band combinations are chosen for analysis of the applied fusion techniques. Also a brief' study about the image fusion quality assessment has been made. Out of them, the quantitative evaluation parameters namely. Peak Signal to Noise Ratio (PSNR), Root Mean Square Error (RMSE) and Correlation Coefficient (CC) are applied using MAI'LAB. It is observed that all the three techniques improved the spatial resolution of the fused Multi- Spectral image but have spectral distortion in varying degrees of extent. The visual comparisons are observed to be matching with the quantitative comparisons up to some degree of extent but are not exactly quantified that up to what terms they are matching. Further the quantitative assessment results are represented graphically for analysis so as to draw conclusions about the suitability of a particular method and its effect on each of the data sets. It is observed that there is no particular method that is suitable for every case. These varying results indicate that every time a single particular quantitative parameter cannot be relied upon for evaluating the quality of the fused images