Controllable-abundance grassland mixed spectrum acquisition experimental method

Abstract

The invention discloses a controllable-abundance grassland mixed spectrum acquisition experimental method. The method includes: measuring the shape of a view field of a spectrograph detector by a 'laser beam interference method'; manually splicing turfs into a grassland mixed pixel in the domain range of view field; simulating natural mixed grasslands different in type, mixing ratio and topological location distribution so as to acquire end element spectrum and mixed spectrum data. The method is applicable to the construction and precision verification of a grassland remote-sensing mixed pixel unmixing model. The method has the advantages that the limitation that the end element spectrum is hard to acquire and unmixing abundances are difficult to verify during unmixing study of natural grasslands is overcome, abundances of the end elements in the grassland mixed spectrum are controllable, mass mixed spectral data with known abundances can be obtained, unmixing accuracy of various mixed spectrum models can be accurately verified, and the method is reliable and has practical value.

Description

technical field [0001] The invention relates to the technical field of hyperspectral quantitative remote sensing applications, in particular to an experimental method for obtaining a grassland mixed spectrum with controllable abundance values, which is used for establishing a grassland mixed pixel decomposition model and evaluating accuracy. Background technique [0002] The remote sensing monitoring of grassland types inevitably has the phenomenon of coexistence of different grassland types in the same pixel. If the mixed pixel problem is not considered, it will inevitably affect the classification accuracy of grassland, and then affect the estimation accuracy of grassland primary productivity. In fact, the mixed pixel Decomposition is also one of the difficulties that must be solved in quantitative remote sensing. Multispectral image classification technology considers that a pixel only contains information of a certain type of ground target, and this pixel is called a pur...

AI Technical Summary

Problems solved by technology

[0004] In the research process of mixed pixel decomposition, to scientifically verify and evaluate various mixed pixel decomposition models, it is necessary to obtain high-precision endmember spectral values ​​and the true value of each endmember abundance value in the pixel. However, in actual operation Due to factors such as spatial resolution, it is difficult to accurately obtain high-precision endmember spectral values ​​from mixed pixels. When the types of ground objects in a pixel are complex, it is also difficult to obtain the true value of each endmember abundance value.

Images