Sparse Tensor Prior for Hyperspectral, Multispectral, and Panchromatic Image Fusion
Xin Tian, Wei Zhang, Dian Yu, Jiayi Ma
Abstract
Dear Editor, This letter is concerned with a new hyperspectral fusion paradigm by simultaneously fusing hyperspectral, multispectral, and panchromatic images. Seeking an efficient prior about the target hyperspectral image (HSI), is vital for constructing an accurate fusion model in this problem. To this end, this work suggests a novel sparse tensor prior using patch-based sparse tensor dictionary learning. It helps improve the effectiveness of the fusion model by fully describing the inherent structures of high-resolution (HR) HSI. By further incorporating this prior with three additional data fidelity terms, the desired fusion model is formulated as an optimization problem and solved efficiently by the alternative direction multiplier method (ADMM). Numerous experimental results are given to validate the effectiveness of the proposed method in terms of visual quality and quantitative analysis.