Fusion refers to combining two or more distinct things, the main objective of employing fusion is to generate results that provides the most detailed, reliable and accurate information possible.  The image fusion is one of the main branchof data fusion. In image fusion the images are fused at different levels of images like pixel, feature and decision level. The necessity of image fusion for high resolution on multispectral and panchromatic images or realtime images for better vision. This paper reviews the general requirements of image fusion, widely used image fusion techniques such as PCA,IHS,DWT,NSCT etc.,summarizes the Quality Assessment Metrics in terms of  metric, description and its principle,  finally image fusion applications in various fields such as object detection, object identification, optimization and pattern reorganization, medical imaging, etc.,

A novel image fusion algorithm based on two filters, one is laplacian filter for de-nosing the detailed coefficients and second filter is Guided Filter (GF) used to refine the approximation as well as detailed coefficient for Computer Tomography (CT) and Magnetic Resonance Imaging (MRI) medical images is proposed. Because of using wavelet transform, we obtained approximation coefficient and other three coefficients of CT and MRI images. Now two weight maps are obtained after the process of denoising. Another reason for obtaining two weight maps is because of comparison. Here comparison is done between two approximation coefficient and six detailed coefficients. By using the approximation coefficients and detailed coefficients, GF is designed. Here GF will guide an image corresponding to the weight maps. Here the weight maps are smoothed using GF and this is mainly served as input image. Hence the weighted fusion algorithm will fuse the both CT and MRI images. A pure fused image is obtained only when the CT and MRI images are refined by inverse wavelet transform. From the comparison results, it can observe that the proposed system gives better results compared to existing system. As well as the proposed system will give maximum amount of input in detail manner

No abstract