矢量数据辅助的高分辨率遥感影像高铁沿线变化检测方法研究

发布时间：2018-03-31 13:55

本文选题：变化检测　切入点：GIS矢量数据　出处：《西南交通大学》2016年硕士论文

【摘要】：利用遥感技术对土地利用资源进行动态监测,已成为各部门了解地区经济活动,进行宏观调控的一种有效手段。随着遥感技术,特别是高空间分辨率的遥感技术以及计算机技术的快速发展,利用该技术对高速铁路沿线进行土地利用检测,从而为铁路部门开展高速铁路沿线环境安全检查提供科学依据和技术支持。因此,本文应用高分辨率遥感影像,就针对高速铁路沿线地物特征以及环境特点进行变化检测方法展开进一步的探讨和分析。本文对遥感影像变化检测相关研究成果进行了归纳总结,研究了基于GIS矢量数据的变化检测方法。当前,基于GIS数据的遥感影像变化检测方法较多,这些方法因变化检测对象的不同而往往具有不同的检测效果,尚没有出现较为成熟的基于GIS数据的遥感影像变化检测框架与体系。因此,本文在总结该类方法总体思路与流程的基础上,详细地介绍了目前几种现有的基于GIS矢量数据的以影像像斑为对象的变化检测方法,同时引入一个新的像斑特征——像斑空间直方图特征,基于该特征提出一种新的基于GIS数据的变化检测方法,通过实验证明该方法在保证效率的同时较直接利用灰度直方图特征进行检测有一定的精度提高。通过实验发现,在第二时相影像中,利用矢量数据分割得到的部分像斑可能会发生局部变化,但在检测过程中将发生局部变化的像斑错误的检测为全部变化或者全部未变化。为此,本文提出了一种通过分析像斑灰度统计分布图来判断第二时期影像像斑的初步变化情况(具体可分为像斑全部变化、像斑全部末变化、像斑局部变化),并通过构建与获取像斑局部变化系数,判断出第二时期发生局部变化的像斑,然后对局部变化的像斑进行再次分割的变化检测方法。实验结果表明,该方法能够检测出利用GIS数据直接分割检测方法不能检测出的部分局部变化区域。检测精度上,在第一组实验数据中,两种利用像斑再分割的检测方法相较于直接利用矢量数据分割结果进行检测分别提高了1.18%、2.69%；在第二组实验数据中,其总体精度分别提高了2.93%、0.54%；在第三组实验数据中,总体精度分别提高了3.36%、3.77%；在第四组实验数据中,总体精度分别提高了2.11%、1.20%。
[Abstract]:Dynamic monitoring of land use resources using remote sensing technology has become an effective means for various departments to understand regional economic activities and carry out macro-control. Especially, with the rapid development of high spatial resolution remote sensing technology and computer technology, land use detection along high-speed railway is carried out. So it provides scientific basis and technical support for railway department to carry out environmental safety inspection along high-speed railway. Therefore, this paper applies high-resolution remote sensing image, This paper further discusses and analyzes the change detection methods for the features of ground objects and environment along the high-speed railway, and summarizes the related research results of remote sensing image change detection. The change detection methods based on GIS vector data are studied. At present, there are many methods based on GIS data for remote sensing image change detection. These methods often have different detection effects because of the different objects of change detection. There has not been a mature framework and system of remote sensing image change detection based on GIS data. This paper introduces in detail several existing methods of image spot detection based on GIS vector data, and introduces a new feature of image spot space histogram. Based on this feature, a new change detection method based on GIS data is proposed. It is proved by experiments that this method has higher accuracy than that of directly using gray histogram features. In the 02:00 phase image, the partial image spots obtained by using vector data segmentation may have local changes, but in the detection process, the local changes will be detected as all or none of the changes. For this reason, In this paper, we propose a method to judge the initial changes of image spots in the second period by analyzing the grayscale distribution of image spots (including all the changes of image spots and all the final changes of image spots). By constructing and obtaining the local variation coefficient of the image spot, the local change of the image spot in the second period is judged, and then the change detection method of the local change image spot is carried out again. The experimental results show that, This method can detect some local change areas which can not be detected by using GIS data directly. In the first set of experimental data, in the first set of experimental data, In the second group of experimental data, the overall accuracy of the two detection methods using image spot re-segmentation was increased by 1.18 and 2.69, respectively, compared with the direct use of vector data. In the third group of experimental data, the overall accuracy was increased by 2.93 and 0.54, respectively. In the third group of experimental data, In the fourth group of experimental data, the overall accuracy was increased by 2.11 and 1.20, respectively.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：P237

【相似文献】