Performance assessment of image fusion algorithms using statistical measures in slant transform domain

Abstract

The important information is collected from multiple input images and formed fused one which has extra quantitative content. Image fusion is executed either in spatial or transform platforms. In spatial domain spectral information is distributed in the entire image. In this work image fusion is implemented in transform domain. Slant Transform effectively represents linear brightness changes. Statistical measure discriminates the important blocks of the image efficiently. Smoothness measure identifies less noisy blocks efficiently. Hence, in this work image fusion in Slant Transform domain using smoothness is proposed. Smoothness of slant transformed blocks are compared to select the important block from multiple images. The eminence of the fused image can be judged using various performance metrics such as Feature Similarity (FSIM) index, Mutual Information (MI), Normalized Cross Correlation (NCC), and Edge Strength Orientation preservation (ESOP).Proposed method is suitable for multi-focus image fusion.