基于GF-2光谱特征的石漠化信息自动提取
发布时间：2019-04-07 19:58
【摘要】：石漠化(Stony Desertification),是指土地处在热带气候、亚热带湿润气候、半湿润气候条件下,以及喀斯特石山区岩溶极其发育的背景下,由于自然因素和人类不合理的经济生产、生活活动等因素导致的土地劣化现象。石漠化是土地退化的一种极端形式,导致缺水、少土和土质贫瘠,石漠化发展最直接的后果就是土地资源的丧失。石漠化与沙漠化、水土流失一起成为我国的三大土地生态灾害。本研究以贵州省贵阳市观山湖区为研究对象,以2016年观山湖区的GF-2遥感数据为基础,通过分析GF-2光谱特征提取观山湖区石漠化小班数据、第二次贵州全省石漠化调查资料及附属资料成果数据、进行实地特征点调查所获取的最新石漠化现状数据,经多次实验验证之后,最终选取了一种能区分GF-2影像中石漠化土地与非石漠化土地的波段运算算法,通过这套算法对GF-2影像进行处理之后,再进行分类。并应用IDL语言编制了一个石漠化信息自动提取模块,应用此模块在ENVI软件中自动提取GF-2遥感影像的石漠化信息。应用此分类方法和自动提取模块,最后获得了贵阳市观山湖区第三次石漠化监测的相关成果,并对观山湖区石漠化进行分级和空间分布特征分析。在此基础上,对观山湖区石漠化监测结果进行了综合分析,得出了观山湖区石漠化的成因与动态变化(与第一次、第二次监测结果进行对比分析)。研究成果可为观山湖区、贵阳市乃至贵州省的石漠化监测和治理提供科学的参考依据。本论文的主要研究成果包括:1)通过对GF-2数据进行0IF指数分析,结合联合熵验证,确定波段2+波段3+波段4组合包含丰富的信息量,是观山湖区石漠化信息提取的最佳波段组合。2)依据贵州省第三次石漠化监测技术标准与GF-2影像的可分性,确定了贵阳市观山湖区石漠化等级划分体系:即非石漠化、潜在石漠化、重度石漠化、中度石漠化、轻度石漠化。3)采用最大似然法对GF-2遥感影像进行分类,分类后的道路与建筑用地界限分明,与真彩色图像契合度极高的同时石漠化斑块的分类效果较好。同时开展面向对象信息提取研究,此时观山湖区的GF-2遥感影像分割合并的参数为分割尺度=47,合并尺度=95。4)分别对最大似然法与基于样本的面向对象石漠化信息提取进行精度评价,其中最大似然法(经过波段运算)的最终分类精度为83.2309%,Kappa系数为0.6752;最大似然法(未经过波段运算)的最终分类精度为77.8021%,Kappa系数为0.6147;基于样本的面向对象石漠化信息提取最终分类精度为74.362%,Kappa系数为0.6233,与。经过波段运算后,最大似然法分类的结果提高了 8.8689个百分点。5)利用IDL语言编写了基于GF-2光谱特征的观山湖区石漠化信息自动提取模块,从而完成了石漠化信息提取的模块化。模块的运行能快速准确地提取石漠化信息,并与ENVI实现有效对接,提高石漠化信息提取的效率。此模块包括波段运算模块与最大似然法模块两个部分。6)制作出了观山湖区石漠化现状和程度分布图,并综合分析了观山湖区石漠化监测区地类分布情况、石漠化与潜在石漠化土地分布情况和石漠化强度分布情况。观山湖区非石漠化面积8165.5hm2,潜在石漠化面积9462.6 hm2,石漠化面积3904.6 hm2。石漠化土地中,轻度石漠化面积1351.8 hm2,中度石漠化面积2148.5 hm2,重度石漠化面积404.3 hm2。
[Abstract]:Stony Destification means that the land is under the condition of tropical climate, subtropical humid climate, semi-humid and humid climate, and in the background that the karst in the karst area is extremely developed, due to the natural factors and the unreasonable economic production of the human, Land degradation caused by factors such as living activities. Rocky desertification is an extreme form of land degradation, resulting in a lack of water, less soil and poor soil, and the most direct consequence of the development of rocky desertification is the loss of land resources. Stone desertification and desertification and water and soil loss become the three major land ecological disasters in China. This study is based on the GF-2 remote sensing data of the Guanyan Lake region of Guiyang, Guizhou Province. Based on the data of GF-2 remote sensing data in the region of Guanshan Lake in 2016, the data of the data and the data on the data of the data and the data on the data of the data of the rock and stone in the whole province of Guizhou Province are analyzed by the analysis of the GF-2 spectrum. The present data of the latest stone in the field characteristic point survey is carried out. After many experiments and verification, a band operation algorithm for distinguishing the land of the stone and the non-stone land in the GF-2 image is finally selected, and after the GF-2 image is processed by the algorithm, the classification is carried out. And an automatic extraction module for stone information is prepared by using the IDL language, and the block information of the GF-2 remote sensing image is automatically extracted in the ENVI software by using the module. The classification method and the automatic extraction module are applied, and the relevant results of the third stone-rock monitoring in the Guanshan Lake region of Guiyang are finally obtained, and the characteristics of the classification and the spatial distribution of the stone-stone in the Guanshan Lake region are analyzed. On the basis of this, a comprehensive analysis of the results of the monitoring of the rock-stone in the Great Lakes region is carried out, and the genetic and dynamic changes of the stone-stone in the Great Lakes region are obtained (compared with the first and the second monitoring results). The research results can provide scientific reference for the monitoring and management of the stone-rock in the Great Lakes region, Guiyang and the Guizhou province. The main research results of this paper are as follows:1) By performing the 0 IF index analysis on the GF-2 data, the combined entropy verification is combined to determine that the band 2 + band 3 + band 4 combination contains a rich amount of information, It is the best wave band combination of the information extraction of the stone-stone in the Guanshan Lake region.2) According to the separability of the third stone-stone monitoring technology standard and the GF-2 image in Guizhou, the division system of the stone-stone level in the Guanyan Lake region of Guiyang is determined: that is, the non-stone-stone, the latent stone-stone and the severe stone-stone. The classification of GF-2 remote sensing image was carried out by using the maximum likelihood method, and the classification of the road and the building was clear, and the classification effect of the stone plaque was better than that of the true color image. At the same time, the object-oriented information extraction is carried out. At this time, the parameters of the GF-2 remote sensing image segmentation and merging in the Guanshan Lake region are the division scale = 47, the combined scale = 95.4), and the accuracy evaluation is carried out on the maximum likelihood method and the sample-based object-oriented stone extraction information respectively, The final classification accuracy of the maximum likelihood method (through band operation) is 83.2309%, the Kappa coefficient is 0.6752, the final classification accuracy of the maximum likelihood method (not through band operation) is 77.8021%, the Kappa coefficient is 0.6147, and the final classification accuracy of the sample-based object-oriented stone extraction information is 74.362%. The Kappa coefficient was 0.6233, compared to. After the wave band operation, the result of the maximum likelihood classification is increased by 8.8689%.5) The automatic extraction module of the stone-level information based on the GF-2 spectral characteristic is prepared by using the IDL language, thus the modularization of the information extraction of the stone information is completed. The operation of the module can extract the stone information quickly and accurately, and can effectively interface with the ENVI to improve the efficiency of the stone extraction information extraction. The module includes two parts of the band operation module and the maximum likelihood module. The distribution of the land and the distribution of the stone and stone. The area of the non-stone area in the Guanshan Lake area is 8165.5hm2, the potential stone area is 9462.6hm2, and the stone area is 3904.6hm2. In the land of Shifang, the area of the light stone is 1351.8hm2, the area of the medium stone is 2148.5hm2, and the area of the severe stone is 404.3hm2.
【学位授予单位】：中南林业科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：X171;X87
，
本文编号：2454377
[Abstract]:Stony Destification means that the land is under the condition of tropical climate, subtropical humid climate, semi-humid and humid climate, and in the background that the karst in the karst area is extremely developed, due to the natural factors and the unreasonable economic production of the human, Land degradation caused by factors such as living activities. Rocky desertification is an extreme form of land degradation, resulting in a lack of water, less soil and poor soil, and the most direct consequence of the development of rocky desertification is the loss of land resources. Stone desertification and desertification and water and soil loss become the three major land ecological disasters in China. This study is based on the GF-2 remote sensing data of the Guanyan Lake region of Guiyang, Guizhou Province. Based on the data of GF-2 remote sensing data in the region of Guanshan Lake in 2016, the data of the data and the data on the data of the data and the data on the data of the data of the rock and stone in the whole province of Guizhou Province are analyzed by the analysis of the GF-2 spectrum. The present data of the latest stone in the field characteristic point survey is carried out. After many experiments and verification, a band operation algorithm for distinguishing the land of the stone and the non-stone land in the GF-2 image is finally selected, and after the GF-2 image is processed by the algorithm, the classification is carried out. And an automatic extraction module for stone information is prepared by using the IDL language, and the block information of the GF-2 remote sensing image is automatically extracted in the ENVI software by using the module. The classification method and the automatic extraction module are applied, and the relevant results of the third stone-rock monitoring in the Guanshan Lake region of Guiyang are finally obtained, and the characteristics of the classification and the spatial distribution of the stone-stone in the Guanshan Lake region are analyzed. On the basis of this, a comprehensive analysis of the results of the monitoring of the rock-stone in the Great Lakes region is carried out, and the genetic and dynamic changes of the stone-stone in the Great Lakes region are obtained (compared with the first and the second monitoring results). The research results can provide scientific reference for the monitoring and management of the stone-rock in the Great Lakes region, Guiyang and the Guizhou province. The main research results of this paper are as follows:1) By performing the 0 IF index analysis on the GF-2 data, the combined entropy verification is combined to determine that the band 2 + band 3 + band 4 combination contains a rich amount of information, It is the best wave band combination of the information extraction of the stone-stone in the Guanshan Lake region.2) According to the separability of the third stone-stone monitoring technology standard and the GF-2 image in Guizhou, the division system of the stone-stone level in the Guanyan Lake region of Guiyang is determined: that is, the non-stone-stone, the latent stone-stone and the severe stone-stone. The classification of GF-2 remote sensing image was carried out by using the maximum likelihood method, and the classification of the road and the building was clear, and the classification effect of the stone plaque was better than that of the true color image. At the same time, the object-oriented information extraction is carried out. At this time, the parameters of the GF-2 remote sensing image segmentation and merging in the Guanshan Lake region are the division scale = 47, the combined scale = 95.4), and the accuracy evaluation is carried out on the maximum likelihood method and the sample-based object-oriented stone extraction information respectively, The final classification accuracy of the maximum likelihood method (through band operation) is 83.2309%, the Kappa coefficient is 0.6752, the final classification accuracy of the maximum likelihood method (not through band operation) is 77.8021%, the Kappa coefficient is 0.6147, and the final classification accuracy of the sample-based object-oriented stone extraction information is 74.362%. The Kappa coefficient was 0.6233, compared to. After the wave band operation, the result of the maximum likelihood classification is increased by 8.8689%.5) The automatic extraction module of the stone-level information based on the GF-2 spectral characteristic is prepared by using the IDL language, thus the modularization of the information extraction of the stone information is completed. The operation of the module can extract the stone information quickly and accurately, and can effectively interface with the ENVI to improve the efficiency of the stone extraction information extraction. The module includes two parts of the band operation module and the maximum likelihood module. The distribution of the land and the distribution of the stone and stone. The area of the non-stone area in the Guanshan Lake area is 8165.5hm2, the potential stone area is 9462.6hm2, and the stone area is 3904.6hm2. In the land of Shifang, the area of the light stone is 1351.8hm2, the area of the medium stone is 2148.5hm2, and the area of the severe stone is 404.3hm2.
【学位授予单位】：中南林业科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：X171;X87
，
本文编号：2454377
本文链接：https://www.wllwen.com/shengtaihuanjingbaohulunwen/2454377.html
