主成分分析技术在遥感蚀变信息提取中的应用研究综述
作者简介:吴志春(1986-),男,江西石城人,硕士,讲师,主要从事遥感地质解译、三维地质建模方面的教学与研究。E-mail: wuzhch_ecit@163.com
收稿日期: 2018-04-18
要求修回日期: 2018-09-04
网络出版日期: 2018-11-20
基金资助
国家自然科学基金(41802247、41603031)
江西省教育厅科技项目(GJJ160584)
放射性地质与勘探技术国防重点学科实验室开放基金项目(RGET1305)
江西省数字国土重点实验室开放基金项目(DLLJ201614)
A Review on Application of Techniques of Principle Component Analysis on Extracting Alteration Information of Remote Sensing
Received date: 2018-04-18
Request revised date: 2018-09-04
Online published: 2018-11-20
Supported by
National Science Foundation of China, No.41802247, 41603031
The Scientific Research Fund of Jiangxi Provincial Education Department, No.GJJ160584
Key Laboratory for Radioactive Geology and Exploration Technology, Fundamental Science for National Defense, No.RGET1305
Key Laboratory for Digital Land and Resources of Jiangxi Province, No.DLLJ201614.
Copyright
主成分分析是目前遥感蚀变异常信息提取常用方法之一,该方法具有对影像大气校正质量要求不高、实现简单、提取效果好、效果稳健等优点,广受地质工作者的青睐。根据输入影像的数量及类型、主成分分析的次数等,本文将主成分分析分为标准主成分分析、特征向量主成分分析、定向主成分分析、二次主成分分析、不同影像间的主成分分析等5种类型。其中,特征向量主成分分析又可细分为4个波段特征向量主成分分析和3个波段特征向量主成分分析。在上述分类的基础上,系统介绍了各种主成分分析及蚀变信息主分量的选择,尤其是对特征向量主成分分析的Crosta技术和定向主成分分析的软落叶技术进行了详细阐述。并以TM/ETM+、ASTER影像为例,对部分应用主成分分析提取蚀变异常信息的实例进行了分析,认为:在基岩裸露区,不同主成分分析都可以很好地提取铁化、泥化蚀变信息;在中、低植被覆盖区,采用标准主成分分析、Crosta技术、改进的Crosta技术、软落叶技术、“掩膜/抑制干扰信息+主成分分析”等方法可以有效地提取蚀变异常信息;高植被覆盖区多采用主成分分析生成的蚀变信息主分量进行彩色合成,再通过对彩色影像进行目视解译的方式判断蚀变的类型和范围。其中,“掩膜+Crosta技术”、“掩膜+软落叶技术”、二次主成分分析等方法在高植被覆盖区也可以取得较好的应用效果;对于干扰信息种类众多、岩性复杂的地区,可根据干扰信息、岩性种类划分成若干个小区,再根据每个小区实际情况采用不同的蚀变提取方法,最后将每个小区内提取的蚀变信息进行合并。
吴志春 , 叶发旺 , 郭福生 , 刘文恒 , 李华亮 , 杨羿 . 主成分分析技术在遥感蚀变信息提取中的应用研究综述[J]. 地球信息科学学报, 2018 , 20(11) : 1644 -1656 . DOI: 10.12082/dqxxkx.2018.180195
The principle component analysis (PCA) technique, as one of the common method of extracting alteration information of remote sensing, is characterized by undemanding quality of atmospheric correction images, easily realization, effective, and steadily i.e., and is widely used by geologists. Based on the number and type of the input images and times of the PCA, this paper subdivided the PCA analysis into the standard principle component analysis (SPCA), feature oriented principal components selection (FPCS), directed principal component analysis (DPCA), the secondary principle component analysis and the principle component analysis of different images, of which the FPCS comprise four bands and three bands principle component analysis. Based on the above mentioned, every PCA and selection criteria have been systematically introduced, especially for the Crosta technique of the FPCS and the software defoliant technique of the DPCA. Images of TM/ETM+、ASTER are selected as examples to analyze the part of application techniques of PCA on extracting alteration information of remote sensing. The results indicate that different PCA all are in favor of extracting information of iron and clay alteration. The methods of SPCA, Crosta technique, modified Crosta technique, software defoliant technique and mask or inhibition of interference information+PCA have been effectively applicated in the medium-low vegetated-covered area. In contrast, in the high vegetated-cover area, the principle component of alteration information derived from PCA was adopted to conduct color composite. The resulted color images were then visual interpreted to estimate the type and extent of alteration. Among these techniques, the “Mask and Crosta technique”, the “Mask and software defoliant technique” and the secondary principle component analysis also can achieve good results in the high vegetated-cover area. For areas characterized by numerous kinds of interference information and complex lithology, firstly, it is practicable to divide the area into several sub-areas based on kinds of interference information and lithology; secondly, different methods of extracting alteration information should be proposed according to the features of every sub-areas; lastly, synthesizing alteration information extracted from every sub-areas.
Fig. 1 Comparision of spatial resolution and spectral resolution of each band of MSS, TM, ETM+ and OIL images图1 MSS、TM、ETM+和OIL各波段空间分辨率、光谱分辨率比较 |
Fig. 2 Mineral spectral curves of iron alteration图2 铁化蚀变矿物光谱曲线 |
Fig. 3 Mineral spectral curves of clay alteration图3 泥化蚀变矿物光谱曲线 |
Tab. 1 Modified Crosta technique表1 改进的Crosta技术 |
技术方法 | 提取的蚀变类型及应用效果 |
---|---|
PCA(TM2,TM4,TM5,TM7) | 提取泥化蚀变和碳酸盐化蚀变[70] |
PCA(TM1,TM4/3,TM5,TM7) | RGB(PC1,PC3,PC4)影像合成,图像中含钾长石斑晶花岗岩和含金钾化硅化蚀变带呈鲜红色,可清楚辨别。通过对PC1和PC4二维散点投图,可将含钾长石斑晶花岗岩、含金钾化硅化蚀变带进一步分离[71]。该方法在加拿大北部冰川、森林覆盖区,中国河北等金矿区得到较好应用[26] |
PCA(TM1+TM2,TM4/3,TM5,TM7) | 干旱基岩裸露区,可以同时提取铁化、泥化蚀变信息,提取的蚀变信息与已知矿床、矿(化)点吻合度高[55,72-73]。在北祁连山西段疏勒河以东约9000 km2范围内,提取的蚀变异常与已知的103个矿床(点)吻合度高达83.5%[55]。在东天山戈壁地区的石英滩至赤湖地区约33 000 km2范围内,提取的蚀变异常与已知的122个矿床(点)吻合度达86%[72] |
PCA(TM1,TM4,TM5/7,TM7) | 在基岩裸露区,运用TM5/7代替TM5,增强了泥化蚀变信息,扩大了泥化蚀变与植被之间的光谱差异[59,74]。吴浩等[59]有效提取了青海省五龙沟金矿区的泥化蚀变信息,提取的蚀变信息在野外得到验证 |
PCA(TM1,TM3/1,TM4,TM5) | 在基岩裸露区,运用TM3/1代替TM3,增强了铁化蚀变信息,扩大了铁化蚀变与植被之间的光谱差异。在青海省五龙沟金矿区提取的铁化蚀变,受断裂构造控制明显,与成矿物质来源和运移通道相吻合,提取的蚀变信息可作为该区的金矿找矿标志[59] |
PCA(TM2,TM3,TM4,TM5/1) | 提取褐铁矿化蚀变异常。提取的蚀变异常信息与金矿体、水系沉积物Au和Cu异常相吻合[70] |
The authors have declared that no competing interests exist.
[1] |
[
|
[2] |
[
|
[3] |
[
|
[4] |
[
|
[5] |
[
|
[6] |
[
|
[7] |
[
|
[8] |
|
[9] |
[
|
[10] |
|
[11] |
[
|
[12] |
|
[13] |
|
[14] |
[
|
[15] |
|
[16] |
|
[17] |
|
[18] |
|
[19] |
|
[20] |
[
|
[21] |
[
|
[22] |
[
|
[23] |
[
|
[24] |
[
|
[25] |
[
|
[26] |
|
[27] |
[
|
[28] |
[
|
[29] |
[
|
[30] |
[
|
[31] |
[
|
[32] |
[
|
[33] |
|
[34] |
|
[35] |
|
[36] |
[
|
[37] |
[
|
[38] |
[
|
[39] |
[
|
[40] |
|
[41] |
|
[42] |
|
[43] |
|
[44] |
[
|
[45] |
[
|
[46] |
|
[47] |
[
|
[48] |
[
|
[49] |
[
|
[50] |
[
|
[51] |
[
|
[52] |
|
[53] |
|
[54] |
[
|
[55] |
[
|
[56] |
[
|
[57] |
[
|
[58] |
|
[59] |
[
|
[60] |
[
|
[61] |
[
|
[62] |
|
[63] |
[
|
[64] |
|
[65] |
|
[66] |
[
|
[67] |
[
|
[68] |
[
|
[69] |
[
|
[70] |
[
|
[71] |
[
|
[72] |
[
|
[73] |
[
|
[74] |
[
|
[75] |
|
[76] |
|
[77] |
|
[78] |
[
|
[79] |
|
[80] |
[
|
[81] |
[
|
[82] |
[
|
[83] |
[
|
[84] |
[
|
[85] |
[
|
[86] |
[
|
[87] |
[
|
[88] |
[
|
/
〈 | 〉 |