Abstract:
In computer vision, image fusion is an eminent and orthodox research topic. Fusion of panchromatic image (PAN) and multispectral image (MS) on pixel level results in high resolution MS image as compared to original MS image. Quite a lot of image fusion methodologies were suggested in literature to make available a high resolution MS image. The main focus is to achieve best quality and more details in the resulting image. Image fusion is trying to get a high quality image from two spatially and spectrally degraded image. The existing image fusion techniques do not provide the desired result. In this thesis report, a fresh way of image fusion is presented that will enhance the spatial and spectral qualities of both images and merged them efficiently. The proposed work apply low pass filter on the PAN image along with edge detection to enhance the edges of PAN image. A contrast based scheme is applied to MS image to enrich spectral quality. The resultant PAN and MS images are fused by R-Fuse method which works with the help of sylvester equation. The suggested technique is implemented in Matlab tool and applied on Deimos2 dataset and Quickbird imagery dataset. The proposed work is assessed by some quality rations such as ERGAS, SAM, SSIM, RMSE, Correlation Coefficient and Universal Image Quality Index. The results shows that the above mentioned work provide enhanced fusion quality as compare to the previous techniques. The superiority of fusion algorithm is confirmed through visual analysis and quantitative assessment.
