基于增强生成对抗网络的滑坡影像集超分辨率重建

针对高分遥感影像用于滑坡智能识别缺少高清训练集的问题,在组建高清滑坡训练集过程中,为充分利用低分辨率滑坡影像,本文采用基于增强型生成对抗网络模型(ESRGAN)实现了低分辨率滑坡影像集超分辨率重建。ESRGAN模型在SRGAN模型基础上,通过移除批归一化层、加入多级残差网络与残差缩放系数,提升了生成器的特征提取性能与稳定性,并采用迁移学习方法,基于毕节滑坡影像集与云南南景高速公路滑坡影像集进行试验验证。试验结果表明,基于迁移学习的ESRGAN模型在峰值信噪比(PSNR)与结构相似性(SSIM)方面获得更高得分,超分辨率重建取得更优结果。本文研究结果为获取滑坡高分辨率遥感影像集提供了一种新的技术方法。     

392BC—2017AD柴达木盆地东北部降水变化及其与太阳活动和强火山喷发的联系

基于采自柴达木盆地内部牦牛山的祁连圆柏树木样芯,新建采样区域过去2710 a的树轮宽度年表,并计算了器测时期（1957—2017 AD）气象数据与树轮宽度年表的相关系数。结果表明：树轮宽度年表与上一年7月至当年6月降水量变化的相关性最高（r=0.753,n=60,p <0.01）,基于此重建了该区392 BC—2017 AD上一年7月至当年6月的降水量变化。器测时期方差解释量高达64%。重建结果显示,在年代际尺度上存在9个湿润期（106—75 BC、6—39 AD、179—229 AD、581—646 AD、823—914 AD、1026—1095 AD、1378—1414 AD、1567—1609 AD和1985—2017 AD）和7个干旱期（328 BC—297 BC、86—151 AD、694—747 AD、1168—1199AD、1444—1525 AD、1680—1725 AD和1792—1860 AD）,1792—1860 AD是过去2400 a最干旱时期。功率谱分析结果表明,过去2400 a柴达木盆地东北部地区降水量存在2～8 a、40 a、80 a和225 a准周...     

1736–2010年华南前汛期始日变化(英文)

The starting dates of the pre-summer rainy season during historical times(1736– 1911) in Fuzhou and Guangzhou of South China, were determined and reconstructed on the basis of historical documents in the Yu-Xue-Fen-Cun archive, together with observed features of precipitation during the pre-summer rainy season. In addition, starting dates of the pre-summer rainy season from 1953 in Fuzhou and from 1952 in Guangzhou were reconstructed for the instrumental period. These data allowed for analyses of inter-annual and inter-decadal changes in the starting dates of the pre-summer rainy season in South China over the past 300 years. Results show that the mean starting date of the pre-summer rainy season in South China was the first pentad of May; in addition, periodicities in the starting dates of 2–3 years, 10 years, and 40 years were detected during the period 1736–1911, and of 2–3 years, 10 years, and 22 years during the instrumental period. From 1736 to 1911, the earliest starting dates at Fuzhou and Guangzhou both occurred at the fourth pentad of April, while the latest starting dates were at the sixth pentad of May in Fuzhou and the first pentad of June in Guangzhou. During the instrumental period, the earliest and latest starting dates were at the fourth pentad of April and the first pentad of June, respectively, in both Fuzhou during 1953–2010 and Guangzhou during 1952–2010. The maximum difference between neighboring decades during 1736–1911 was 2.2 and 1.6 pentads in Fuzhou and Guangzhou, respectively, and during the instrumental period it was 2.5 and 2.4 pentads in Fuzhou and Guangzhou, respectively.     

土地利用/覆被历史重建方法(英文)

Understanding long-term human-environment interactions requires historical reconstruction of past land-use and land-cover changes. Most reconstructions have been based primarily on consistently available and relatively standardized information from historical sources. Based on available data sources and a retrospective research, in this paper we review the approaches and methods of the digital reconstruction and analyze their advantages and possible constraints in the following aspects:(1) Historical documents contain qualitative or semi-quantitative information about past land use, which also usually include land-cover data, but preparation of archival documents is very time-consuming.(2) Historical maps and pictures offer visual and spatial quantitative land-cover information.(3) Natural archive has significant advantages as a method for reconstructing past vegetation and has its unique possibilities especially when historical records are missing or lacking, but it has great limits of rebuilding certain land-cover types.(4) Historical reconstruction models have been gradually developed from empirical models to mechanistic ones. The method does not only reconstruct the quantity of land use/cover in historical periods, but it also reproduces the spatial distribution. Yet there are still few historical land-cover datasets with high spatial resolution.(5) Reconstruction method based on multiple-source data and multidisciplinary research could build historical land-cover from multiple perspectives, complement the missing data, verify reconstruction results and thus improve reconstruction accuracy. However, there are challenges that make the method still in the exploratory stage. This method can be a long-term development goal for the historical land-cover reconstruction. Researchers should focus on rebuilding historical land-cover dataset with high spatial resolution by developing new models so that the study results could be effectively applied in simulations of climatic and ecological effects.     

Context-based automatic reconstruction and texturing of 3D urban terrain for quick-response tasks

Highly detailed 3D urban terrain models are the base for quick response tasks with indispensable human participation, e.g., disaster management. Thus, it is important to automate and accelerate the process of urban terrain modeling from sensor data such that the resulting 3D model is semantic, compact, recognizable, and easily usable for training and simulation purposes. To provide essential geometric attributes, buildings and trees must be identified among elevated objects in digital surface models. After building ground-plan estimation and roof details analysis, images from oblique airborne imagery are used to cover building faces with up-to-date texture thus achieving a better recognizability of the model. The three steps of the texturing procedure are sensor pose estimation, assessment of polygons projected into the images, and texture synthesis. Free geographic data, providing additional information about streets, forest areas, and other topographic object types, suppress false alarms and enrich the reconstruction results.     

基于无人机遥感影像拓扑分析的三维重建

在无需任何地面控制点或其它先验知识前提下,探索一种基于无人机遥感影像的三维重建方法.利用无人机飞控数据建立的影像拓扑结构,依次通过特征点提取、影像匹配、从运动恢复结构等步骤估计出相机位置和姿态参数,并恢复出场景特征点云信息,最后对重建精度进行分析.试验结果表明,文中提出的方法可快速、可靠地实现较高精度的三维模型重建.     

基于InSAR技术对西部高山地区DEM的重建

合成孔径雷达干涉测量技术是一种获取地面数字高程模型(DEM)的新技术.介绍利用InSAR获取DEM的基本流程,阐述相位解缠的理论方法以及基线估计中获取地面控制点的方法与原则,通过实验获取西部某高山地区的DEM,并与SRTM DEM作比较、分析,总结出其存在局限性的原因.     

一种改进的Mean Shift点云数据滤波

点云数据滤波是三维重建质量好坏的关键,为了避免在进行滤波时花费大量计算时间建立点云间的拓扑关系,提出利用改进的Mean Shift算法直接对点云数据进行处理,使其快速移动到核密度估计函数的最大值点,从而达到降低噪声的目的。     

广义点摄影测量在建筑物立面轮廓重构中的应用

采用广义点摄影测量理论对建筑物立面框架轮廓进行重构。广义点是对当代线摄影测量以及无穷远点（灭点）理论的升华。实验结果表明,采用线面基元进行立面重构相比于经典的点云重构方法,不仅简洁、高效,还能获得清晰的三维边缘轮廓,且自动化程度较高。     

基于临汾2次强震史料与GIS技术对未来强震灾害预测评估

通过对临汾2次历史强震史料的研究,总结出其震害分布特征,对比各类衰减模型绘制的等震线图,确定了对该地区不同震级地震最适用的衰减关系模型和修正值;应用GIS技术的空间分析功能,实现了临汾地区更为直观的震害预测评估系统.再将临汾2次历史强震损失重建,从而预测出临汾地区现有经济社会情况下的地震灾害的损失情况,为其未来防震减灾...