基于高分辨率遥感影像的典型地震次生地质灾害快速智能提取研究

发布时间：2018-07-11 16:02

本文选题：遥感 + 地震次生地质灾害　；参考：《中国科学院大学(中国科学院遥感与数字地球研究所)》2017年硕士论文

【摘要】：地震次生地质灾害具有形成时间短、突发性强、破坏性大、诱导性强等特点,能够加剧地震破坏效应,严重阻碍灾后应急救援工作的顺利实施。快速获取地震次生地质灾害灾情信息,对震后救援、应急以及重建等工作具有十分重要的意义。遥感技术以其快速、宏观、经济等优势,逐步成为空间对地观测的重要手段之一。传统的基于遥感技术的震后次生灾害调查,多是通过对遥感数据人工目视解译结合辅助地理数据综合分析判定完成的。随着遥感技术、空间技术、计算机技术的飞速发展,基于遥感技术的地震次生灾害快速识别和检测逐步向自动化、智能化方向发展。灾后高分辨率遥感影像的精细化分类、多时相遥感数据的变化检测以及综合利用影像多特征分析等人机交互方式,已逐步成为当前遥感技术应用于灾害应急的主要手段,也是近年来国内外学者研究的热点之一。地震次生地质灾害在高分辨率遥感影像上的光谱、形态、纹理、空间位置分布等特征,是次生地质灾害遥感信息提取的物理基础。充分把握地震次生地质灾害的多种遥感特征,选择合理的特征参量,实现对灾害信息的快速、智能提取是高分辨率遥感信息提取的研究方向之一。此外,从计算机视觉角度,利用视觉显著性分析理论,快速提取显著目标区域进而得到特定的目标信息是计算机视觉领域的研究热点之一。通过对视觉注意模型的分析和应用,针对地震次生地质灾害在遥感影像上的显著性表现,可以有效地识别和检测次生地质灾害。本论文分别从经典的多特征分析和视觉显著性分析两方面对地震次生地质灾害遥感信息快速、智能提取进行研究,主要取得了以下成果:(1)提出了一种基于震后高分辨率遥感影像的地震次生地质灾害自动提取方法——基于多特征的层次分析方法。该方法综合利用了灾害影像的光谱、形状、纹理及空间分布等特征,通过选择合理的特征参量、自动化地计算参量阈值、利用专家知识建立合理的提取规则,实现了对地震次生地质灾害的快速、智能提取。该方法的最大优势在于快速、智能,不依赖于人机交互,对地震灾害应急救援具有很大的现实意义。(2)提出了两种结合显著性分析和超像素分割的典型地震次生地质灾害快速提取方法。两种方法均以显著性检测为基础,通过对原始影像做超像素分割实现处理对象从单像元级到超像素级的扩展。两种方法分别利用统计学习和超像素区域相似性度量方法实现了对显著目标的提取。相关实验对灾后滑坡体和泥石流进行了提取,以解决在复杂多目标的遥感图像上显著性检测能力不足问题。实验证明,该算法对具有较为明显的且类型一致或差异不大的显著目标提取效果较好。由于该类方法以显著性检测为基础,对于遥感影像中显著性较低的水体提取效果不佳,因此不太适合对灾后堰塞湖信息进行提取。(3)将视觉显著性检测方法与经典的多特征分析方法对比分析,提出一种仅使用震后单时相高分辨率遥感影像的地震次生地质灾害遥感信息快速、智能提取新模式。论文的创新点主要有以下几个方面:1)提出了基于多特征综合的层次分析方法对地震次生地质灾害进行提取;2)通过改进的Otsu算法,实现对特征参量阈值的自动计算;3)提出两种结合显著性检测和超像素分割的遥感信息提取方法,并成功应用到地震次生地质灾害的遥感信息提取中;4)将视觉显著性检测与多特征综合分析方法进行对比,提出一种针对地震次生地质灾害遥感信息快速、智能提取的新模式。
[Abstract]:The secondary geological disaster of earthquake has the characteristics of short formation time, strong sudden, destructive and strong inducibility, which can aggravate the earthquake damage effect and seriously impede the successful implementation of emergency rescue work after the disaster. It is of great significance for the rescue, emergency and reconstruction after the earthquake to obtain the disaster information of the secondary geological disasters. Remote sensing technology, with its rapid, macro and economic advantages, has gradually become one of the important means of space observation. The traditional post earthquake disaster investigation based on remote sensing technology is determined by the synthetic analysis of artificial visual interpretation of remote sensing data combined with the comprehensive analysis of auxiliary geographic data. With remote sensing, space technology and computer technology Rapid development of the rapid identification and detection of secondary earthquake disaster based on remote sensing technology has gradually developed into automation and intelligent direction. The fine classification of high resolution remote sensing images after disaster, the change detection of multi-phase remote sensing data and the comprehensive utilization of multi feature analysis of images have gradually become the current remote sensing technology. The main means used for disaster emergency is also one of the hot spots of scholars at home and abroad in recent years. The characteristics of the secondary geological disasters on high resolution remote sensing images are the physical basis of the extraction of remote sensing information of secondary geological disasters. Feature, selecting reasonable characteristic parameters to realize fast disaster information, intelligent extraction is one of the research directions of high resolution remote sensing information extraction. In addition, from the angle of computer vision, using visual significance analysis theory, fast extraction of significant target area and then get specific target information is the research heat of computer vision field. One point. Through the analysis and application of the visual attention model, the secondary geological disasters can be effectively identified and detected in view of the significant performance of the secondary geological hazards on the remote sensing images. This paper is fast and intelligent from the classical multi feature analysis and visual significance analysis to the two sides of the earthquake secondary geological hazard remote sensing information. The main results are as follows: (1) a method of automatic extraction of earthquake secondary geological hazards based on high resolution remote sensing images after the earthquake is proposed, which is based on the multi feature analytic hierarchy process. This method uses the characteristics of the spectrum, shape, texture and spatial distribution of the disaster images, by selecting reasonable special features. Parameters, automatically calculate parameter threshold, use expert knowledge to establish reasonable extraction rules, realize rapid and intelligent extraction of earthquake secondary geological disasters. The greatest advantage of this method is fast, intelligent, not dependent on human computer interaction, and has great practical significance for earthquake disaster emergency aid assistance. (2) two kinds of combination display are put forward. The two methods are based on the saliency detection of the typical seismic secondary geological hazards. The two methods use statistical learning and the similarity measure of super pixel region, respectively, on the basis of the super pixel segmentation of the original image. In order to solve the problem of lack of significant detection ability in remote sensing images with complex multi-target, the experimental results show that the algorithm has a better effect on the extraction of significant targets with more obvious types of uniform type or small difference. On the basis of the significance detection, it is not very suitable for extracting the information of the dammed lake after the disaster. (3) to compare the visual significance detection method with the classical multi feature analysis method, and propose a kind of seismic time which only uses the single phase high-resolution remote sensing image after the earthquake. The new pattern of remote sensing information of geological disasters is fast and intelligent. The main innovations of this paper are as follows: 1) a hierarchical analysis method based on multi feature synthesis is proposed to extract the earthquake secondary geological disasters; 2) the automatic calculation of the threshold value of the characteristic parameter is realized by the improved Otsu algorithm; 3) two kinds of combined saliency detection are put forward. The remote sensing information extraction method of measuring and super pixel segmentation has been successfully applied to the extraction of remote sensing information of earthquake secondary geological disasters. 4) comparing the visual significance detection with the multi feature comprehensive analysis method, a new model for rapid and intelligent extraction of remote sensing information of earthquake secondary geological disasters is proposed.
【学位授予单位】：中国科学院大学(中国科学院遥感与数字地球研究所)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P642.27;TP79

【相似文献】