| Launch：2019-07-26 |
Abstract
The new method for extracting spectral information (end-element) from hyperspectral remote sensing images in combination with spatial information of images. It is an effective way to understand the interior information of the image pixels more accurately
Inrtoduction
Recently, Kong Xiangbing, an engineer from the Institute of Soil and Water Conservation of the Yellow River Institute of Water Conservancy, worked with Professor Shuning of Wuhan University to propose a method for extracting spectral information (end-element) from hyperspectral remote sensing images in combination with spatial information of images. It is an effective way to understand the interior information of the image pixels more accurately. Related papers were published in the Journal of 《spectroscopy and spectral analysis》.
Hyperspectral remote sensing uses many very narrow electromagnetic wave bands to detect objects far from the target and non-contact objects, and to obtain the reflected, radiated or scattered electromagnetic wave information. Because of the limited spatial resolution of the image and the complexity of the object distribution in the objective world, many important objects are only hidden in the hyperspectral remote sensing image pixels. How to accurately "mine" the information of these sub-pixels is the key to the quantitative application of hyperspectral remote sensing.
It is found that the distribution of terrain has a certain distance correlation in space, and the closer the spatial distance is, the greater the similarity of terrain is; the distribution of terrain in different regions has its unique characteristics, with the distribution law of "large mixed, small settlements". Based on the above understanding, adding image spatial information in each stage of sub-pixel recognition process can significantly improve the accuracy of spectral information extraction.
