HySpex影像地形校正与降维技术研究
文献类型：学位论文
| 作者 | 吴梦娟 |
| 学位类别 | 硕士 |
| 答辩日期 | 2016 |
| 授予单位 | 中国科学院大学 |
| 授予地点 | 北京 |
| 导师 | 周可法 |
| 关键词 | HySpex影像 正射校正 地形辐射校正 数据降维 |
| 学位专业 | 地图学与地理信息系统 |
| 中文摘要 | 随着高光谱遥感及其数据处理技术的发展，成像高光谱技术逐渐成为遥感应用的前沿。成像高光谱遥感最大的特点就是“图谱合一”，即在获取地表物质光谱特征的同时能够得到该物体的图像特征，所拍摄影像拥有上百个波段，光谱分辨率可达到纳米级，能够识别地表更加细微的光谱特征，实现对地表物质更加精细的识别与分析。机载成像高光谱技术在对地表物质进行探测时具有明显优势，但是地形起伏、数据量大等问题都会影响后续的信息识别与提取，进而影响到应用效果。因此，本文将主要对地形起伏引起的几何变形、辐射畸变，以及高光谱数据的维数聚联问题展开研究。本文采用搭载于动力三角翼平台上的HySpex成像高光谱仪拍摄的高光谱影像。在进行遥感影像采集过程中，动力三角翼飞行高度较低（一般在30-150m），地形变化对遥感影像数据影响很大，数据应用之前需要对HySpex影像进行两个方面的地形校正，即基于地形的正射校正和基于地形的辐射校正，前者主要校正图像的几何形变，后者校正影像的辐射亮度。文中采用的HySpex全谱段高光谱成像仪VNIR-1600、SWIR-384共有504个波段，为了降低数据冗余并且改善信息提取和后续分类效果，需要对数据进行降维处理。本论文的研究内容与结果主要有：（1）研究分析现有的正射校正模型，并基于HySpex数据的特征，采用有理函数模型，对HySpex数据进行地形几何校正，调整校正中的DEM分辨率，观察不同DEM分辨率下的地形校正效果，并将结果进行对比。结果表明，并不是DEM分辨率越大地形校正效果越好，并且降低DEM分辨率可以提高计算效率。（2）依据HySpex数据的特征，优化筛选合适的地形辐射校正模型，将处理结果进行评价对比分析，最终得出在地形辐射校正中，改进C校正处理效果较好，并且效率较高（3）优化筛选适合HySpex数据的降维方法，将降维结果进行评价对比，结果表明KPCA在HySpex数据降维中效果最好，不仅能保留较多的数据特征，并且降维后的波段数明显减少。 |
| 英文摘要 | With the development of remote sensing and remote sensing data processing technology, hyperspectral imaging technology is becoming frontier in remote sensing applications. The greatest feature of the imaging hyperspectral remote sensing is the combination of image and spectrum, we could obtain the spectral characteristics of surface material while the image feature of the object. Because of hyperspectral remote sensing imaging having the number of multi-band that the spectral resolution can achieve nanometer, it is possible to identify more subtle spectral characteristics of the surface, so as to achieve the identification and analysis of surface material. Although airborne hyperspectral imaging technology can give full play to its advantages when it was used to detect the surface material, due to the interference of topography, data volume and other issues affected the identification and subsequent information extraction. Therefore, this paper focuses on discussing and solving the geometric and radiation distortion caused by topography, and the curse of dimensionality of hyperspectral data. The data in this paper is acquired by the HySpex imaging spectrometer mounted on the delta wing platform. Since the delta wing flying very low height of only 30-150m, so the remote sensing image acquisition is greatly affected by the terrain. The terrain correction of the acquired HySpex image is divided into two areas, the orthorectification and radiation based on terrain. The former is mainly corrected image geometric distortion, the latter is the radiance corrected image. This paper is adopted HySpex full spectrum hyperspectral imager VNIR-1600, SWIR-384 of its 504 total number of bands, it is necessary to reduce the dimension data. Dimensional reduction technique is premising of high-spectral image information extraction and subsequent classification. The main research contents and results are as follows: (1) Analyze existing orthorectification model, and based on the characteristics of HySpex data, selected the rational function model for terrain geometric correction of HySpex data. Adjusting the DEM resolution in topographic correction, this paper observed the effects under the different DEM resolution, and compared the results. Eventually, this paper can draw a conclusion that the greater DEM resolution can improve computation efficiency. (2) Based on the characteristics of HySpex data, selected the suitable terrain radiant correction model. The results will be evaluated and comparative analysis, the final results improved that the improved C correction is better and more efficient. (3) Dimensionality reduction method for optimizing screening HySpex data, the dimension reduction results were evaluated. And, the results show that the KPCA is the optimal in the HySpex data dimensional reduction while it is not only retaining more data features, but also obviously reducing the number of bands after reduced dimensionality. |
| 学科主题 | 地图学与地理信息系统 |
| 语种 | 中文 |
| 源URL | [http://ir.xjlas.org/handle/365004/15000]  |
| 专题 | 新疆生态与地理研究所_研究系统_荒漠环境研究室 |
| 作者单位 | 中科院新疆生态与地理研究所 |
| 推荐引用方式 GB/T 7714 | 吴梦娟. HySpex影像地形校正与降维技术研究[D]. 北京. 中国科学院大学. 2016. |
入库方式： OAI收割
来源：新疆生态与地理研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。