基于成像光谱数据的岩矿信息识别系统设计与实现

【作者】 侯军；

【导师】 张宏；

【作者基本信息】 南京理工大学 ， 计算机应用技术， 2012， 硕士

【摘要】 成像光谱数据的光谱分辨率很高,包含很多从可见光到热红外波段的光谱信息,是一个高维的光谱数据集,同时又包含丰富的空间、辐射信息,给地质勘探应用带来了新的机遇。基于成像光谱数据的岩矿信息识别是遥感地质勘探领域的一个研究热点。本文面向地质矿产勘探,围绕基于成像光谱数据的典型蚀变矿物和岩性信息识别展开研究。首先,在总结成像光谱遥感探测技术特点、分析岩矿识别技术研究现状的基础上,面向成像光谱数据设计了一种半监督的典型蚀变矿物与岩性信息提取与识别方法,给出了典型岩矿信息识别的业务化处理流程；其次,进行成像光谱数据岩矿信息识别系统框架,以及岩矿信息识别、岩矿分类填图等核心模块的详细设计；然后,在系统框架和功能模块详细设计的基础上进行系统的开发与实现,包括成像光谱数据图像端元提取和光谱匹配识别等关键步骤以及最小噪声分离、纯净像元指数、光谱匹配等核心算法的研究与实现；最后,基于实际航空成像光谱数据进行系统试验分析,验证了本文岩矿信息识别系统的有效性,可用于辅助地质矿产勘探的实践工作。更多还原

【Abstract】 Spectrographic data has the very high spectral resolution, which can collect spectrum information simultaneously in hundreds of narrow and contiguously spaced bands, with wavelengths ranging from the visible spectrum to the infrared region. Thanks to the high dimensional data and the rich space, radiation information, spectrographic remote sensing has been increasingly utilized for mineral exploration. The information identification of rock and mineral in spectrographic data is a research hotspot in the field of remote sensing in geology.The identification information of the typical altered rock and mineral is studied for mineral exploraion. First of all, based on the characteristics of hyperspctral remote sensing detection technology and the present research of rock and mineral identification, it designs a half supervision of the typical altered minerals and litholoy information extraction and identification methods of spectrographic data, and gives the processing of the typical rock and mineral identification; Secondly, it designs the system framwork of information identification of rock and mineral in spectrographic data and the core modules of information identification, classification and mapping of rock and mineral in detail; Then, it explores and designs this system based on the detailed design of the system framwork and function modules, including the key steps, such as endmember exraction, spectrum matching and identification of spectrographic data, and gives the detailed design of the important algorithms, such as minimum noise fraction, pure pixel index, spectrum angle matching; At last, it tests the actual spectrographic data systemly, and the result shows that the information identification system of rock and mineral is available, and can be used to help the parctice work of mineral exploration.更多还原