作者: Xuejian Sun , Lifu Zhang , Hang Yang , Taixia Wu , Yi Cen
DOI: 10.1109/JSTARS.2014.2356512
关键词: Computer science 、 Remote sensing 、 Transformation matrix 、 Hyperspectral imaging 、 Pixel 、 Spectral resolution 、 Image resolution 、 Matrix (mathematics) 、 Multispectral image 、 Set (abstract data type)
摘要: Hyperspectral (HS) remote sensing has an important role in a wide variety of fields. However, its rapid progress been constrained due to the narrow swath HS images. This paper proposes spectral resolution enhancement method (SREM) for remotely sensed multispectral (MS) image, generate images using auxiliary multi/hyper-spectral data. Firstly, set number spectra different materials are extracted from both MS and Secondly, approach makes use linear relationships between multi hyper-spectra specific transformation matrices. Then, angle weighted minimum distance (SAWMD) matching is used select suitable matrix create vectors original pixel by pixel. The final result image data same as that spatial were also derived matrices can be multitemporal periods. was tested with three datasets, spectra-enhanced real compared visual interpretation, statistical analysis, classification evaluate performance. experimental results demonstrated SREM produces good data, which will not only greatly improve range applications but encourage more utilization