基于IDL的青海湖流域草地分类及其生物量监测遥感系统开发与应用

发布时间：2018-03-23 13:11

本文选题：草地类型　切入点：草地生物量　出处：《山东农业大学》2015年硕士论文

【摘要】：本文针对遥感数据处理的特点和图像可视化的相关特性,选取IDL作为“青海湖流域草地分类及其生物量监测遥感系统”的开发语言。针对系统的功能需求和实际开发重点,对如何利用IDL草地遥感数据的相关处理、数据的可视化分析和应用等进行了详细的设计,对系统的设计思路、技术路线和实现方法进行了详细的论述。结合IDL开发平台软件设计技术,通过开发实例,在系统用户交互性、可视化效果和扩展性方面进行尝试。最终取得以下成果:(1)文章分析比较多种编程语言及平台,选择经济、便捷的ENVI+IDL二次开发技术搭建系统。研究了IDL语言的特点,重点深入研究各种关键技术,分别了设计了主控模块、草地遥感分类模块、植被指数计算模块、草地生物量计算模块。系统能够通过提取分析遥感数据,实现草地快速分类,计算草地物量、覆盖度、叶面积指数等草地参数,达到对草地快速监测的目的。(2)运用了马氏距离分类、欧氏距离分类、光谱角填图法、最大似然分类、决策树分类法对草地进行遥感分类。其中决策树和最大似然法的分类结果明显优于其它方法。其中最大似然法的分类精度最高,达到77.8%。分析了6个植被指数NDVI、RVI、DVI、EVI、MSAVI和SAVI与草地地上生物量之间均存在着不同程度的相关性;其中,RVI与生物量之间的相关性最高(相关系数0.776)。比较以各种光谱指数为自变量建立的线性、对数、二次曲线和三次曲线回归模型。通过分析比较,最后确定以RVI为自变量的三次曲线模型y=3.9852x3-17.661x2+70.785x+65.624精度最高,R2达到0.687,是青海湖环湖地区草地生物量监测的最佳植被指数模型。(3)系统搭建完成后,对青海湖流域的草地类型及生物量进行了计算,并实现计算结果的快速可视化显示。本系统具有友好的人机交互界面,操作简易。本系统集成了在草地研究中的常用的研究算法,有大量的提示信息,使研究人员不需复杂的专业软件即可对草地遥感图像快速处理。系统各功能模块均能良好快速实现,包括草地遥感分类模块、植被指数计算模块、草地生物量计算模块等主要功能的实现。系统可视化效果较为理想。对影像做了2%线性拉伸,使得影像整体色调适中,符合人眼的视觉要求。可实现对图像平移、缩放、鹰眼预览、鼠标取值等操作,方便图像的浏览系统可扩展性强。系统各模块都独立开发,减少了相互之间的依赖,使模块功能可依需要随时删减。系统优化合理。编程时使用函数来代替运行效率慢的循环语句,及时释放内存中的是失效的变量,优化内存占用量。系统运行速度比较令人满意。
[Abstract]:According to the characteristics of remote sensing data processing and the related characteristics of image visualization, this paper selects IDL as the development language of "remote sensing system for grassland classification and biomass monitoring in Qinghai Lake Basin". How to use IDL grassland remote sensing data related processing, data visual analysis and application are designed in detail. The technical route and implementation method are discussed in detail. Combined with the software design technology of IDL development platform, through the development of examples, the system user interaction, Finally, this paper analyzes and compares many programming languages and platforms, and chooses the economical and convenient secondary development technology of ENVI IDL to build a system. The characteristics of IDL language are studied. The main control module, grassland remote sensing classification module, vegetation index calculation module and grassland biomass calculation module are designed respectively. The system can extract and analyze remote sensing data to realize the rapid classification of grassland. The grassland parameters, such as grassland quantity, coverage, leaf area index and so on, were calculated to achieve the purpose of rapid monitoring of grassland. (2) Markov distance classification, Euclidean distance classification, spectral angle mapping method, maximum likelihood classification were used. Decision tree classification method is used to classify grassland by remote sensing. The result of decision tree and maximum likelihood method is obviously superior to other methods, and the maximum likelihood method has the highest classification accuracy. The correlation between the six vegetation indices (NDVI RVI VIVI and SAVI) and the aboveground biomass of grassland were analyzed. The correlation between RVI and biomass is the highest (correlation coefficient 0.776). The regression models of linear, logarithmic, conic and cubic curves, which are established by using various spectral indices as independent variables, are compared. Finally, the cubic curve model (y=3.9852x3-17.661x2 70.785x 65.624) with RVI as independent variable was determined to be the best vegetation index model for monitoring grassland biomass around Qinghai Lake, and the R2 was 0.687, which was the best vegetation index model for monitoring grassland biomass around Qinghai Lake. The grassland types and biomass in Qinghai Lake basin are calculated, and the results are visualized quickly. The system has friendly man-machine interface. Easy to operate. This system integrates the common research algorithms in grassland research, and has a lot of information. So that researchers do not need complex professional software to quickly process remote sensing images of grassland. All functional modules of the system can be realized quickly and well, including remote sensing classification module of grassland, vegetation index calculation module, The realization of the main functions such as biomass calculation module of grassland. The visualization effect of the system is relatively ideal. The image is stretched by 2% linear, which makes the overall color of the image moderate and meets the visual requirements of the human eye. The system can realize the translation and scaling of the image. Eagle eye preview, mouse values and other operations, convenient image browsing system is extensible. Each module of the system is independently developed, reducing mutual dependence, The module function can be deleted at any time according to the need. The system optimization is reasonable. In programming, the function is used to replace the slow cycle statement, and the invalid variable is released in the memory in time. Optimized memory usage. System running speed is satisfactory.
【学位授予单位】：山东农业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：S812;S818.9

【参考文献】