The recital assessment of image fusion based on multi-resolution transforms using smoothness approach proach

Abstract

Combining the information from the registered source images is the process involved in the image-fusion. In this present manuscript, two fusion rules are explored. The first one is based on the weightage-based rule. The second one is the Smoothness and weightage-based algorithm. Smoothness is used to reduce the noise from the source images. These two methods are independent of the selection of the transform. In this work, Discrete Wavelet Transform is considered to perform the experimentation. The recital comparison was made between multiresolution transforms using maximum selection method, weightage method, and smoothness along with weightage. The source images are generally multi-focused, satellite, panchromatic, and clinical medical images. The experimental results show that more smoothened (in addition with weightage) images (including edges and curves) provide high visual information. The advantage of this approach is proven using the performance metrics such as PSNR, NCC, MI, ESOP, and FSIM. The blocking artifacts are reduced by decomposing the transforms and the high frequency noise in the source image is smoothened by proposed approaches.