Landsat8 OLI图像增强与岩性识别方法

新疆维吾尔自治区阿克陶县布伦口地区海拔高、穿越条件差、岩性复杂,因而研究Landsat8 OLI数据信息增强与岩性识别在该区有强大的应用需求,也有遥感岩性识别的理论意义。区内不同岩性单元光谱较为相似,仅利用最佳波段组合(OIF)指数、主成分分析(PCA)和独立主成分分析(ICA)等图像增强方法均不能最大限度地区分各岩性单元。通过分析独立主成分变换之后各岩性单元的特征向量,发现不同岩性单元特征向量差异比原图像光谱差异增大,为扩大差异,依据特征向量统计结果构建比值运算,再联合PCA特征向量进行彩色合成,各岩性单元可识别性显著增加。研究表明:(1)采用联合多种图像增强的方法能将研究区不同岩性单元进行分离;(2)国道314南北两侧地层均划归布伦阔勒岩群的结论有误,其北侧仍属布伦阔岩石,南侧实为片麻状二长花岗岩;(3)沙湖南东原划归志留纪的岩石单元可进一步分解为志留纪地层和元古代布伦阔勒岩群岩石单元。该方法有推广应用价值和深化研究的前景。

An image enhancement and lithology identification method based on Landsat8 OLI data

The lithology is complex in the Bulunkou area, Aketao county, Xinjiang, where the altitude is high and the traffic conditions are poor. The research on information enhancement and lithology recognition using Landsat8 OLI data has practical application demand as well as theoretical significance in identification of remote sensing lithology. Using the optimum index factor (OIF) band combination, PCA (principal component analysis) and ICA (independent principal component analysis) transform image enhancement cannot distinguish between different lithologic units completely because the spectral curves of lithologic units are similar in the research area. Based on the analysis of the lithologic unit vector characteristics after independent principal component transformation, we found that the difference enlargement between different lithologic unit characteristic vectors after ICA transform and the original image spectrum characteristic difference. In order to further expand the differences, ratio calculation was performed based on the statistics results of characteristic vectors, and the color composite image was built with PCA feature vectors and ratio images. In the composite images, identification of all lithologic units can be increased significantly, permitting to divide the rock units on the basis of the final synthesis image in the study area. The research shows that: (1) Combining a variety of image enhancement methods can separate different lithologic units. (2) The lithology units on both sides of the 314 Road in the east of the Sha lake that are classified as Bulunkuo lithology unit are not incorrect. We find that the lithology unit north of the 314 Road is still Bulunkuo rock unit, however, and that in the south of the road is gneissoid monzonitic granite. (3) The rock unit in the southeast of the Shahu lake that was classified as silurian can be decomposed into silurian strata and proterozoic bulunkuo rocks. This method can be applied to a broader range and deserves further research.