多时相高分辨率震害遥感影像建筑物变化检测研究
发布时间：2018-12-26 18:34
【摘要】：随着遥感技术突飞猛进的发展，遥感数据的获取呈现出多平台、多传感器、多分辨率（多空间分辨率、多光谱分辨率、多时间分辨率）的新阶段。由于数据种类不断增多，质量不断提高，基于高分辨率遥感影像的变化检测研究也日趋频繁，，应用领域甚广。目前各种破坏性极强的自然灾害频繁发生，造成巨大的财产损失和人员伤亡。利用多时相高分辨遥感影像进行灾害分析也成为时下研究热点，本文主要研究地震后建筑物的变化状况，该项研究将会为震后应急救援、灾后重建工作发挥重要作用，具体研究内容体现在以下几点： 1）本文首先对高分辨率遥感影像的特性进行了分析总结，找出了适合本研究的影像方面的相关特性，并对常用遥感变化检测方法的原理、优缺点进行了详细分析和阐述，包括影像算术运算法、植被指数差值法、主成分分析法以及分类后比较法。 2）面向对象的对象分类技术研究，包括多尺度影像分割原理和影像模糊分类方法。 3）提出了一种逐级多特征方向的变化检测方法，该方法的主要步骤是：首先进行像元级变化检测，获取包含建筑物变化在内的大的候选区域；其次在大候选区域上进行纹理特征的变化检测；最后在纹理特征检测的基础上进行色调特征的变化检测，这样逐级进行不但缩小了变化检测范围，还降低了其他地物的干扰。为了提高该方法的效率，在进行变化检测之前必须进行数据预处理工作，本文也详细介绍了辐射校正和几何精校正的方法和原理。 4）类间叠置分析统计法，利用了面向对象的影像处理技术和矢量数据空间分析中叠置分析的方法，即首先通过影像分类把建筑物类提取出来，再利用叠置分析的方法找出受损的建筑物区域，并根据相关系数对建筑物受损状况进行了评估，得到损失程度信息。 最后本文对2010年青海省玉树藏族自治州玉树县发生的7.1级大地震的部分受灾区域的数据进行了实验，实验结果表明，本文提出的两种方法均有效提高了建筑物变化检测的正确率，降低了漏检和虚检的情况。
[Abstract]:With the rapid development of remote sensing technology, the acquisition of remote sensing data presents a new stage of multi-platform, multi-sensor, multi-resolution (multi-spatial resolution, multi-spectral resolution, multi-time resolution). Because of the increasing variety of data and the improvement of quality, the research on change detection based on high resolution remote sensing image is becoming more and more frequent, and the application field is very wide. At present, a variety of destructive natural disasters occur frequently, resulting in huge property losses and casualties. Disaster analysis using multitemporal high-resolution remote sensing images has become a hot topic. This paper mainly studies the changes of buildings after earthquake, which will play an important role in post-earthquake emergency rescue and post-disaster reconstruction. The specific research contents are as follows: 1) the characteristics of high-resolution remote sensing images are analyzed and summarized in this paper, and the relevant characteristics of the images suitable for this study are found out, and the principle of common remote sensing change detection methods is also discussed. The advantages and disadvantages are analyzed and expounded in detail, including image arithmetic method, vegetation index difference method, principal component analysis method and post-classification comparison method. 2) object-oriented object classification technology, including multi-scale image segmentation principle and image fuzzy classification method. 3) A change detection method is proposed in this paper. The main steps of this method are as follows: firstly, pixel level change detection is carried out to obtain large candidate areas including building changes; Secondly, the change detection of texture feature is carried out on the large candidate region. Finally, the change detection of tone feature is carried out on the basis of texture feature detection, which not only reduces the range of change detection, but also reduces the interference of other ground objects. In order to improve the efficiency of the method, data preprocessing must be carried out before the change detection. The methods and principles of radiation correction and geometric precision correction are also introduced in detail in this paper. 4) the statistical method of inter-class superposition analysis, which uses the object-oriented image processing technology and the method of overlay analysis in vector data space analysis, that is, the class of buildings is extracted by image classification. Then the damaged building area is found by the method of overlay analysis, and the damage degree information is obtained by evaluating the damage condition of the building according to the correlation coefficient. Finally, the data of the area affected by the magnitude 7.1 earthquake occurred in Yushu County, Yushu Autonomous Prefecture, Qinghai Province, in 2010 have been tested. The experimental results show that, The two methods proposed in this paper can effectively improve the correct rate of building change detection and reduce the situation of missed and false detection.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P237
本文编号：2392524
[Abstract]:With the rapid development of remote sensing technology, the acquisition of remote sensing data presents a new stage of multi-platform, multi-sensor, multi-resolution (multi-spatial resolution, multi-spectral resolution, multi-time resolution). Because of the increasing variety of data and the improvement of quality, the research on change detection based on high resolution remote sensing image is becoming more and more frequent, and the application field is very wide. At present, a variety of destructive natural disasters occur frequently, resulting in huge property losses and casualties. Disaster analysis using multitemporal high-resolution remote sensing images has become a hot topic. This paper mainly studies the changes of buildings after earthquake, which will play an important role in post-earthquake emergency rescue and post-disaster reconstruction. The specific research contents are as follows: 1) the characteristics of high-resolution remote sensing images are analyzed and summarized in this paper, and the relevant characteristics of the images suitable for this study are found out, and the principle of common remote sensing change detection methods is also discussed. The advantages and disadvantages are analyzed and expounded in detail, including image arithmetic method, vegetation index difference method, principal component analysis method and post-classification comparison method. 2) object-oriented object classification technology, including multi-scale image segmentation principle and image fuzzy classification method. 3) A change detection method is proposed in this paper. The main steps of this method are as follows: firstly, pixel level change detection is carried out to obtain large candidate areas including building changes; Secondly, the change detection of texture feature is carried out on the large candidate region. Finally, the change detection of tone feature is carried out on the basis of texture feature detection, which not only reduces the range of change detection, but also reduces the interference of other ground objects. In order to improve the efficiency of the method, data preprocessing must be carried out before the change detection. The methods and principles of radiation correction and geometric precision correction are also introduced in detail in this paper. 4) the statistical method of inter-class superposition analysis, which uses the object-oriented image processing technology and the method of overlay analysis in vector data space analysis, that is, the class of buildings is extracted by image classification. Then the damaged building area is found by the method of overlay analysis, and the damage degree information is obtained by evaluating the damage condition of the building according to the correlation coefficient. Finally, the data of the area affected by the magnitude 7.1 earthquake occurred in Yushu County, Yushu Autonomous Prefecture, Qinghai Province, in 2010 have been tested. The experimental results show that, The two methods proposed in this paper can effectively improve the correct rate of building change detection and reduce the situation of missed and false detection.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P237
【参考文献】
相关博士学位论文 前1条
1 孟磊;采煤驱动下平原小流域生态演变规律及评价[D];中国矿业大学;2010年
本文编号：2392524
本文链接：https://www.wllwen.com/kejilunwen/dizhicehuilunwen/2392524.html
