MAP Based Super-Resolution Method for Hyperspectral Imagery

Hyperspectral imagery (HSI) is used in more and more fields, but its low spatial resolution limits its applications severely. The super-resolution algorithm catches more and more eyes but has not been solved well. In this case, the present paper aimed to do the following researches. The relation modeling was constructed between observed HSI of low resolution and target HSI of high resolution. In the modeling, space transformation was implemented by introducing the operator related to endmembers (EMs) of interest. Maximum posterior probability (MAP) algorithm was used to realize the super-resolution (SR) recovery. Experiments show that the proposed SR method has good recovery effect, low computational complexity, robust noise resistance, and can preserve classes of interest.