An Improved Image Fusion Technique based on Wavelet and Edge Gradients to Augment Spatial Prominence of Image Features

Abstract

Image Fusion techniques being motivated by the fact to register and preserve images with perceptually meaningful features. Recently, transform based techniques are found to be competent and severely being used to fuse or combine images. In this paper, an improved wavelet & edge based technique known as edgelets to improve the spatial prominence of image features in image fusion is proposed. Firstly, images to be fused are transformed into YCbCr color space and obtained edges using gradient technique. Wavelet is applied to individual color channles (Y, Cb, Cr) with adaptive weights at different subbands. Weights are iteratively updated at all decomposition levels for better convergence of efficiency in fused image during reconstruction. Proposed method is proved dominant using standard metrics such as PFE, PSNR, Corr, SD & SF in comparison with very recent state-of-the-art techniques.