A Novel Approach to Preserve Small Scale Details in Fused Image using Guide Filter with NSCT for Visual and Infrared Images

Abstract

Fusion of more than a single image create a new version of image which holds plenty of information producing suitable and accurate decision making for the real-world applications. There is no accurate fusion mechanism to produce such a complement detailed image even though huge number of techniques and methodologies are available in the literature. This is an attempt to analyze the pros and cons of the existing popular fusion methods and provide a methodology to contribute in a fusion community. In this approach, the source images are fused by NSCT method. Though NSCT is good in preserving small scale details, it introduces some artifacts in the fused image. To overcome this drawback, the residual of the sources are derived as they retain small scale details such as edges and textures. The fusion version of the source-residuals is bind with NSCT fused image, an approach to compensate the loss details in this way. The resulted fusion images are compared with other techniques qualitatively and quantitatively. Result shows that the proposed approach outperforms the other approaches.