Local Principal Component Averaging Image Fusion

Abstract

Image fusion plays a vital role to enhance the perception of images by integrating information of source images without introducing any form of degradations. Various fields such as medical imaging, remote sensing, night vision applications and surveillance find wide applications of image fusion methods not only to enhance the image details but also in denoising like post processing techniques. In this paper, modified principal component analysis, named, local principal component averaging fusion (LPCAF) is proposed to integrate multifocus, multisource and day-night images. Fusion rule is proposed based on average of principal components from local regions and performance analysis is carried out with other fusion algorithms based on mutual information, quality index and average peak signal to noise ratio. Simulation and analysis clearly show that the proposed algorithm proves better than other well known fusion techniques.