LiDAR结合高分辨率影像的城市不透水地表提取研究

发布时间：2018-07-26 10:59

【摘要】：城市化带来的大规模不透水地表扩展己对城市生态环境造成重要影响,使原本透水性较好自然资源变成不具有透水性的建设用地,城市不透水地表的增加会加剧城市水资源的污染和植被的减少,使城市面临着严重的生态环境问题。城市不透水地表不仅仅是衡量城市化发展程度的重要指示器,还是衡量城市环境变化与社会经济发展的关键技术指标之一。准确有效地提取城市不透水地表信息,对城市可持续发展与规划有着一定的指导作用。如何从环境复杂的城市地物中分类出详细地物以及提取出不透水地表,是当前遥感领域的热点难点之一。本文以提取城区的不透水地表为最终目的,将湖南省衡阳市某开发区作为研究区,用机载LiDAR数据结合高分辨率影像数据协同联合处理,采用面向对象方法,分别对该城区的不透水地表信息提取使用单一影像数据和多源数据相对应的分类方法得到分类结果,并对其进行对比评价分析,完成精度较高的城市地物类型详细分类。实现城市不透水地表信息的精细分类提取,为高精度不透水地表遥感估算提供了一种新的思路与方法。本文的创新性有:结合LiDAR与同机高分辨率航空影像,完成对城市地物信息的详细分类;在仅有RGB影像的限制条件下,采用新型算法完成对不透水地表相关地物类型的提取;结合LiDAR与高分辨率航空影像的各自优点,完成对较易混淆地物类型的甄别。本文提出的地物分割尺度与分类方法体系,对于该类影像可提供有效的分类参考。得出的主要结论是:(1)对面向对象分析技术进行介绍,详细论述了对影像的分割算法和模糊数学分类方法。主要对多尺度分割算法的理论、方法进行了较为完善的概述,对分割参数的选定方法进行了详细的论述。其次,重点介绍了模糊分类理论,并在此基础上阐述模糊分类规则及分类体系等。(2)构建对研究区的R、G、B高分辨率航空影像的监督分类工作流程与分类体系。多尺度分割是面向对象技术的关键及分类前提,故本文在所有分类步骤之前先对影像进行多尺度分割,找到适用于城市各地物类型的分割参数,最后根据监督分类规则与方法,得到城区的详细地物的分类结果以及不透水地表。(3)建立一套针对城市不透水地表信息提取的技术流程、分类体系及规则。综合LiDAR数据和同机获取的航空影像两种数据源提取城区的不透水地表,采用模糊分类方法,创建一种适用于城市不透水地表的提取方法。分别用最佳分类结果和混淆矩阵精度验证方法对结果进行评价分析,结果表明,该方法可得到较满意的城区复杂地物的不透水地表信息。(4)多源数据协同方法可提高影像分类精度。通过单一航空影像和影像与LiDAR数据联合这两种形式提取城市不透水地表,对比二者的分类结果精度评价,发现多源数据在提取不透水地表时可通过数据的优势互补,使得分类结果最优化。同时利用二者数据特征结合的方法,对可分性较差的地物,如裸地等此类特殊地物提出详细合适的分类方法与特征参考,并得到较为精确的分类结果。
[Abstract]:The large-scale impermeable surface expansion brought about by urbanization has an important impact on the urban ecological environment, which makes the original water permeable and good natural resources become non permeable construction land. The increase of urban water permeable surface will aggravate the pollution of urban water resources and reduce the vegetation, and make the city face serious ecological environment problems. Urban water permeable surface is not only an important indicator to measure the degree of urbanization, but also one of the key technical indicators to measure urban environmental change and social and economic development. Accurate and effective extraction of urban water permeable surface information has a certain guiding role for urban sustainable development and planning. It is one of the hot and difficult points in the field of remote sensing to classify the detailed objects and extract the impermeable surface. In this paper, a development area in Hengyang, Hunan province is taken as the research area, and the airborne LiDAR data combined with high resolution image data is combined together, and the object oriented method is adopted. The classification results of the unpermeable surface information of the urban area are extracted and used in the classification of the single image data and the multi source data, and the comparative evaluation and analysis are carried out to complete the detailed classification of the urban land types with high precision. The fine classification and extraction of the information of the urban water permeable surface is realized, and the high precision and the water surface is not permeable to the surface. Remote sensing estimation provides a new idea and method. The innovation of this paper is: combining LiDAR and high resolution aerial images to complete the detailed classification of urban ground information. Under the limited conditions of RGB image, a new algorithm is used to complete the extraction of the type of the surface related to the surface of the impervious surface; combined with LiDAR and high resolution aviation The main conclusions of this paper are as follows: (1) the main conclusions are as follows: (1) the object oriented analysis technology is introduced, and the image segmentation algorithm and fuzzy mathematical classification are discussed in detail. Methods. The theory and method of multi scale segmentation algorithm are summarized, and the selected methods of the segmentation parameters are discussed in detail. Secondly, the fuzzy classification theory is introduced, and the fuzzy classification rules and classification systems are expounded on this basis. (2) the high resolution aerial images of R, G and B in the research area are constructed. Supervised classification workflow and classification system. Multi-scale segmentation is the key of object oriented technology and classification premise, so this paper divides the image into multi scale before all the classification steps, and finds the segmentation parameters suitable for various types of city objects. Finally, according to the supervised classification rules and methods, the detailed classification of the urban area is obtained. The results and impermeable surface. (3) establish a set of technical process, classification system and rules for urban water permeable surface information extraction, integrated LiDAR data and two kinds of data sources obtained by the same machine to extract the water surface of the urban area, and use the fuzzy classification method to create a method for extracting the water surface of the city. The results are evaluated and analyzed by the best classification results and the obfuscation matrix accuracy verification method. The results show that the method can obtain more satisfactory surface information of the complex terrain in urban areas. (4) the multi source data coordination method can improve the image classification accuracy. The two forms are combined with the single aerial image and the image and LiDAR data. Taking the urban impervious surface and evaluating the accuracy of the classification results of the two, it is found that the multi source data can be optimized by the advantages of the data in the extraction of the impervious surface and make the classification results optimized. At the same time, using the method of combining the characteristics of the two data, the detailed and appropriate points are put forward for the special objects such as the bare land, such as the bare land. Class method and feature reference, and get more accurate classification results.
【学位授予单位】：新疆大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P237

【参考文献】