基于空间遥感影像的居民地自动解译

针对当前地物要素信息采集劳动强度大、智能化程度不高、效率低等技术瓶颈,本文以深度学习理论为基础,在Caffe框架上依托Digits网络服务器构建居民地数据集进行分类识别训练,建立样本数据集并完成模型训练工作.整合居民地复杂数据信息,设计了针对遥感影像自动解译居民地的操作流程.通过对实验结果分析,利用深度学习获取的数据模...     

APPLICATION OF DEM GENERATION TECHNOLOGY FROM HIGH RESOLUTION SATELLITE IMAGE IN QUANTITATIVE ACTIVE TECTONICS STUDY: A CASE STUDY OF FAULT SCARPS IN THE SOUTHERN MARGIN OF KUMISHI BASIN

Traditional method to generate Digital Elevation Model (DEM)through topographic map and topographic measurement has weak points such as low efficiency, long operating time and small range. The emergence of DEM-generation technology from high resolution satellite image provides a new method for rapid acquisition of large terrain and geomorphic data, which greatly improves the efficiency of data acquisition. This method costs lower compared with LiDAR (Light Detection and Ranging), has large coverage compared with SfM (Structure from Motion). However, there is still lack of report on whether the accuracy of DEM generated from stereo-imagery satisfies the quantitative research of active tectonics. This research is based on LPS (Leica Photogrammetry Suit)software platform, using Worldview-2 panchromatic stereo-imagery as data source, selecting Kumishi Basin in eastern Tianshan Mountains with little vegetation as study area. We generated 0.5m resolution DEM of 5-km swath along the newly discovered rupture zone at the south of Kumishi Basin, measured the height of fault scarps on different levels of alluvial fans based on the DEM, then compared with the scarp height measured by differential GPS survey in the field to analyze the accuracy of the extracted DEM. The results show that the elevation difference between the topographic profiles derived from the extracted DEM and surveyed by differential GPS ranges from -2.82 to 4.87m. The shape of the fault scarp can be finely depicted and the deviation is 0.30m after elevation correction. The accuracy of measuring the height of fault scarps can reach 0.22m, which meets the need of high-precision quantitative research of active tectonics. It provides great convenience for rapidly obtaining fine geometry, profiles morphology, vertical dislocations of fault and important reference for sites selection for trench excavation, slip rate, and samples. This method has broad prospects in the study of active tectonics.     

THE EXTRACTION OF HOUSE DISTRIBUTION BASED ON PHOTOGRAMMETRY METHOD:TAKING THE COUNTRYSIDE IN THE WEST OF CHINA FOR EXAMPLE

The key parameters of houses such as distribution, area and height play an important role for urban-rural planning, earthquake emergency and disaster mitigation. The computer automatic extraction method is an effective way to acquire large area house information using satellite-borne or airborne optical remote-sensing images. However, because of the similarity of spectral characters for different land cover types or the influence of snow coverage, the classification accuracy of house type using traditional spectral based method can be decreased. To acquire the accurate houses distribution, a method based on the height information is proposed using unmanned aerial vehicle(UAV)in this study. With UAV flying at the height of 100m above ground, the route of the UVA was planned with the heading direction overlap of 77% and side direction overlap of 50%for the nearby pictures. Taking Qionghalajun Village in Xinjiang Uygur Autonomous Region for example, 69 pictures of the study area were obtained with DJI Phantom 3 professional. With those pictures input into the EasyUAV software, the Digital Elevation Model(DEM), Digital Surface Model(DSM), and Digital Orthophoto Map(DOM)were acquired based on photogrammetry method using the overlapped optical remote-sensing images of UAV. After that, the house distribution and height were acquired with the differences between DSM and DEM images larger than 2.6m. To eliminate the influences of disintegrated pixels on the house extraction, mainly caused by the trees or noise point, the classification aggregation tool of ENVI software was used with the disintegrated pixels' area less than 4m². Compared with visual interpretation result, the user accuracy and mapping accuracy of the house extraction method proposed in this study is 88.69% and 97.42%, respectively. In addition, to evaluate the performance of the proposed method, the result of traditional supervised classification method using DOM data acquired previously was compared with the result of new method. The results show that the new method is more accurate the user accuracy and mapping accuracy of the supervised classification method, which is 43.23% and 85.30%, respectively. Besides the study area in this study, the performance of the proposed method will be evaluated at the other places in the further study.     

利用NOAA/AVHRR资料监测南方林区森林火灾的研究

本文论述了南方林区森林火灾时间分布集中、发生率高、面积小,并以地表火为主的特征;总结了气象卫星资料的林火信息特征和非林火干扰信息规律;研究了建造林火监测专家系统的理论方法,进而提出了林火宏观实时监测系统的总体设计,并建立了以计算机图像处理为基础的、以林火背景数据库支持的、以专家系统为核心的林火监测系统。并利用南方林火模拟实验和历史林火资料对该监测系统进行了改进。最后讨论了林火监测系统改进方向和研究体会。     

空-谱角匹配与多指数法相结合的OLI遥感影像水体提取

针对复杂环境条件下水体遥感提取结果不连续且易与植被、建筑物、阴影相混淆的难题,基于Landsat 8 OLI影像,以石家庄市平山县岗南水库和宿迁市骆马湖附近河流为研究区,提出了一种空-谱角匹配与多指数法相结合的水体信息提取方法;并与单波段阈值法、归一化差分水体指数法（NDWI）、光谱角匹配法（SAM）、自动水体提取指数法（AWEI）和一类支持向量机法（OC-SVM）的水体提取结果进行对比分析和精度评定。试验结果表明,本文提出的方法兼顾了多特征之间的互补性优势,引入的空间信息有效地抑制了噪声的干扰,且以像素为基元的提取策略较好地保持了水体的边缘信息,避免了出现平滑掉细节信息的情况;与传统方法相比,本文方法受植被、建筑和阴影的干扰最小,对细小水体也具备较好的识别能力。     

一种适用于航空影像的无参考模糊探测方法

针对侧风、强风、湍流等飞行环境容易造成航空影像运动模糊,严重影响航空影像质量,同时航空影像数据量大,手动挑选模糊影像费时费力的问题,为了提高航摄内业人员的工作效率,该文研究一种适用于航空影像的自动模糊探测方法,以主流的无参考再模糊算法Reblur和无参考结构清晰度算法NRSS为基础,结合航空影像具有丰富地物的特点,对影像进行分块处理,计算所有字块的Reblur和NRSS模糊探测值,最后得到整幅影像的模糊探测值。其中,再模糊算法通过计算待测影像和参考影像的水平和垂直运动方向上的灰度变化来评价图像模糊度;NRSS算法在结构相似度SSIM算法基础上加入梯度信息提取和高斯滤波等改进,通过计算结构相似度评价图像模糊度。实验结果表明,该文研究的无参考模糊影像探测方法适用于航空影像数据,其评价结果与人眼主观评价结果具有较高的一致性,能够准确地缩小模糊影像的查找范围,极大地提高了航摄内业效率。     

Modelling relational contexts in GEOBIA framework for improving urban land-cover mapping

In human cognition, both visual features (i.e., spectrum, geometry and texture) and relational contexts (i.e. spatial relations) are used to interpret very-high-resolution (VHR) images. However, most existing classification methods only consider visual features, thus classification performances are susceptible to the confusion of visual features and the complexity of geographic objects in VHR images. On the contrary, relational contexts between geographic objects are some kinds of spatial knowledge, thus they can help to correct initial classification errors in a classification post-processing. This study presents the models for formalizing relational contexts, including relative relations (like alongness, betweeness, among, and surrounding), direction relation (azimuth) and their combination. The formalized relational contexts were further used to define locally contextual regions to identify those objects that should be reclassified in a post-classification process and to improve the results of an initial classification. The experimental results demonstrate that the relational contexts can significantly improve the accuracies of buildings, water, trees, roads, other surfaces and shadows. The relational contexts as well as their combinations can be regarded as a contribution to post-processing classification techniques in GEOBIA framework, and help to recognize image objects that cannot be distinguished in an initial classification.     

基于对象BOW特征的高分辨率遥感影像变化检测方法

针对传统基于像素的变化检测方法的缺点,以及底层特征表现能力不足等问题,提出一种基于对象BOW特征的变化检测方法。首先,将经过预处理操作的两期影像进行波段组合得到组合后影像,再考虑地物光谱特征和几何空间信息对组合后影像进行多尺度分割,获得相对应的对象基元;同时,分别提取两幅影像的底层特征（包括影像各波段的均值和方差以及灰度图像的6种纹理特征）。其次,将对象视作文档,像素的特征向量视作单词,利用BOW模型构建影像对象的中层表达,即对象的BOW特征。最后,通过相似性度量算法比较相应对象的BOW特征,从而识别出影像上的变化区域。本文利用2组WorldView-2影像进行了检验,结果表明本文方法的变化检测结果较为完整,精度优于对比方法。本文方法基本能够满足变化检测的需求,为高分辨率遥感影像上的数据挖掘分析提供了有效的手段。     

A Bayesian characterization of urban land use configurations from VHR remote sensing images

The composition and arrangement of spatial entities, i.e., land cover objects, play a key role in distinguishing land use types from very high resolution (VHR) remote sensing images, in particular in urban environments. This paper presents a new method to characterize the spatial arrangement for urban land use extraction using VHR images. We derive an adjacency unit matrix to represent the spatial arrangement of land cover objects obtained from a VHR image, and use a graph convolutional network to quantify the spatial arrangement by extracting hidden features from adjacency unit matrices. The distribution of the spatial arrangement variables, i.e., hidden features, and the spatial composition variables, i.e., widely used land use indicators, are then estimated. We use a Bayesian method to integrate the variables of spatial arrangement and composition for urban land use extraction. Experiments were conducted using three VHR images acquired in two urban areas: a Pleiades image in Wuhan in 2013, a Superview image in Wuhan in 2019, and a GeoEye image in Oklahoma City in 2012. Our results show that the proposed method provides an effective means to characterize the spatial arrangement of land cover objects, and produces urban land use extractions with overall accuracies (i.e., 86% and 93%) higher than existing methods (i.e., 83% and 88%) that use spatial arrangement information based on building types on the Pleiades and GeoEye datasets. Moreover, it is unnecessary to further categorize the dominant land cover type into finer types for the characterization of spatial arrangement. We conclude that the proposed method has a high potential for the characterization of urban structure using different VHR images, and for the extraction of urban land use in different urban areas.