Research On Spectral Analysis And Its Application To Mineralization Information Extraction

The spectral characteristics of ground objects are theoretical basis of remote sensing applications. The technology methods and processes of mineral alteration information extraction are usually chosen based on the spectral reflection characteristics of field measured data. In fact, the mineral alteration information always performs very "weak" in remote sensing image because it is always a "small categories" target in the objects under research. Additionally, the atmosphere, mix of objects and other factors affect the spectral reflection characteristics of the remote sensing image data. So, it is not enough only using the spectral reflection characters of field measured data to guide the mineral alteration information extraction in remote sensing.The spectral characteristics of field measured data and ETM image data in Tonren-Zeku area, Huangnan state, Qinghai Province are comparatively analyzed in spectral curve space and spectral feature space. With the focus set on the clustering structure of the objects in the spectral feature space, and the spectral aberration of remote sensing data caused by atmosphere and the mix of objects, as well as its solutions. Then, the model of mineral alteration information extraction is built based on the results of the spectral characteristics research. This model was used in the mineral alteration information extraction of Awa area, Huangnan state, and achieved good results.The main contents of the resarch include the following:(1) The spectral reflection characteristics of typical objects in the spectral curve space are analyzed and summarized, with the results showing that:the same ground features have similar spectral reflectance curves, and the spectral reflectance curves of heterogeneous ground features have more obvious differences, which indicate that the spectrums of different objects are absolutely separable in a physical sense.(2) Research on the distribution shape and relationship of the typical objects in spectral feature space, with the results showing that:the spectrum of field data is evidently separable in the statistical space, however, the separability of the spectrum of ETM remote sensing data is decreased because of the atmospheric disturbance and the mixture and other factors.(3) Research on the spectral aberrance of ETM remote sensing data caused by atmospheric disturbance and the mix of objects, and the method for remedy.(4) The model of mineral alteration information extraction is built based on the results of the spectral characteristics research. This model is used in the mineral alteration information extraction of Awa area, Huangnan state, with the results showing high precision and reliability.The results of the pilot study of this paper provide more scientific basis for weak mineral alteration information extracting. This paper proposes a comparative research on geometric lattice structure of field measuered data and remote sensing data in spectral feature space, and using the spatial structure of field measuered data to guide the analysis and application of remote sensing data, which provides a new research approach for the researchers who study on the classification or information extraction with remote sensing data.