基于多光谱图像的高光谱图像模拟分析.docx

Abstract
To meet the needs of a variety of different remote sensing tasks, such as agriculture, forestry, geology, marine, meteorology, hydrology, military, environmental protection and other fields, people usually want to be able to get remote sensing images of different spaces, spectrums that reflect the variety of natural conditions, under different remote sensing platforms. Due to the impact of atmo- spheric conditions and sensor observations, remote sensing images on the above conditions are not available in any case. In fact, the purpose can be achieved by image simulation technology.
The simulation of remote sensing image has been one of the important research directions of the remote munity, which has been a great development. Simulations of remote sensing images are able to alternate real remote sensing images under a variety of conditions, meeting the needs of different remote sensing research. Firstly, the paper analyzes the current remote sensing image simulation methods and then introduces the typical spectrum databases at home and abroad. On this basis, the paper proposes a method of optical remote sensing image simulation, which uses multispectral image to simulate hyper-spectral image. Firstly, pre-processings are taken on the input multi-spectral image, including atmospheric correction and radiometric correction and so on, to achieve surface reflectance image; the next step is to extract from reflectance image thus pose mixed pixel into pure end-member spectrums and abundance map .Thirdly, interpolating the end-member spectrums with the help of spectrum databases, In the end, mix the subdivision spectrum and the abundance into simulated hyper-spectrum. In addition, the proposed method is to identify collection correct feature type from the interpolated spectrum which is the subdivision spectrum that has more narrow bands. In this way, we can improve recognition pared to the direct use of multispectral identification. This method is used in the sim