空间插值算法研究及其在遥感数据模拟中的应用

发布时间：2018-04-11 11:27

本文选题：空间插值 + 克里金插值算法　；参考：《成都理工大学》2013年硕士论文

【摘要】：随着社会经济的不断发展，科学技术的不断进步，使得人类对空间环境的作用越来越强。一方面，人们对于资源地质信息预测和分析的精度要求越来越高，范围要求越来越广；另一方面，矿产预测与新兴科学技术的联系越来越紧密，并不断有新方法、新技术的提出，特别是遥感技术的快速发展，传递给我们的信息越来越多，人们也越来越认识到信息技术对资源获取所起到的重大作用。另一方面，有时会因为多方面的原因以至于不能获取完整有效的遥感数据。利用临近的已知空间数据对未知空间数据值进行估计和推测，是解决缺省或无效空间数据非常有效的手段，即空间插值。本文详细介绍了常用空间插值算法，在克里金（kriging）插值算法和分形插值算法研究的基础上，将以上两种算法应用于适当调整之后的高光谱遥感数据。主要工作包括以下几个方面：（1）简要介绍了空间常用插值算法理论；（2）克里金插值算法的研究；（3）分形插值算法的研究；（4）利用克里金插值和分形插值算法对高光谱遥感影像数据缺省波段进行插值运算，并通过峰值信噪比等图像信息参量对试验结果进行验证评估。本文的主要创新点有：（1）因为克里金插值算法和分形插值算法虽然在地质领域运用很广，但是运用于高光谱遥感影像光谱曲线插值还非常少，本文通过对高光谱遥感数据立方体进行空间坐标分解之后，运用克里金插值和分形插值对其假设缺省波段进行插值模拟；（2）在应用插值算法的时候，为了保证算法精度的无偏性和平稳性，本文将波段数据中空值坐标区域全部换为波段均值，在插值模拟完成之后，，再将对应坐标位置的象元值替换为0，保证和原始数据空间结构一致。论文最后给出了克里金插值和分形插值对文中高光谱数据插值的对比结果，包括克里金多种模型和多个波段的实验结果。实验结果表明，除极少数跳跃波段外，本文中所应用的克里金（kriging）插值和分形插值算法对高光谱影像数据具有较好的模拟效果，具有一定的应用价值，并对论文工作进行了总结，对空间插值算法进行了深入的分析与展望。
[Abstract]:With the development of social economy and the progress of science and technology, human beings play a more and more important role in space environment.On the one hand, the accuracy and scope of prediction and analysis of resource geological information are becoming more and more demanding; on the other hand, the relationship between mineral prediction and emerging science and technology is becoming closer and closer, and new methods are constantly available.With the development of new technology, especially the rapid development of remote sensing technology, more and more information has been transmitted to us, and more and more people realize that information technology plays an important role in resource acquisition.On the other hand, remote sensing data can not be obtained for many reasons.The estimation and estimation of unknown spatial data using adjacent known spatial data is a very effective method to solve the default or invalid spatial data, namely spatial interpolation.Based on the research of Kriging interpolation algorithm and fractal interpolation algorithm, the above two algorithms are applied to the hyperspectral remote sensing data after proper adjustment.The main tasks include the following:Firstly, the theory of spatial interpolation algorithm is introduced briefly.2) the research of Kriging interpolation algorithm;3) Fractal interpolation algorithm;(4) Kriging interpolation and fractal interpolation are used to interpolate the default band of hyperspectral remote sensing image data, and the experimental results are verified and evaluated by image information parameters such as peak signal-to-noise ratio (PSNR).The main innovations of this paper are:Because Kriging interpolation algorithm and fractal interpolation algorithm are widely used in geological field, but they are still very few in hyperspectral remote sensing image spectral curve interpolation.After the hyperspectral remote sensing data cube is decomposed into spatial coordinates, the Kriging interpolation and fractal interpolation are used to simulate the hypothetical default band of hyperspectral remote sensing data cube.In order to ensure the accuracy of the interpolation algorithm, in order to ensure the accuracy of the algorithm unbiased and stationary, this paper changes the band data hollow value coordinate region to the band mean value, and after the interpolation simulation is completed,Then the pixel value of the corresponding coordinate position is replaced with 0, which is consistent with the original data spatial structure.At the end of this paper, the comparison between Kriging interpolation and fractal interpolation for hyperspectral data interpolation is given, including the experimental results of several Kriging models and multiple bands.The experimental results show that, except for a few jump bands, the Kriging Kriging-based interpolation and fractal interpolation algorithms used in this paper have a good simulation effect on hyperspectral image data, and have certain application value, and the work of this paper is summarized.The spatial interpolation algorithm is analyzed and prospected.
【学位授予单位】：成都理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P237

【参考文献】