融合空间和时序遥感信息的深度学习水稻提取

为了得到更加精细的水稻提取结果,提出一种结合高分辨率和多时序遥感影像的深度学习水稻提取方法.构建全卷积网络(FCN)对BJ-2高分辨率遥感影像进行分类,并利用长短期记忆网络(LSTM)和随机森林(RF)分类器对Sentinel-2多时序遥感影像进行分类,再通过面向对象的分割和投票对3种方法的分类结果进行融合,得到最终提...     

自由波核章动的光学和新技术检测

本文应用当今最高精度的经典仪器光学观测资料、新技术的综合观测资料和单一的ＶＬＢＩ观测资料检测了自由液核章动，得到其周期为４１５—４１８天，分析了其运动形态为逆向的圆周运动，并计算了其振幅为亚毫秒级．     

Tri-frequency spectrum method and results for resolving the parameters of Earth’s liquid core free nutation

The parameters of Earth free core nutation (FCN) are two relatively significant geophysical parameters. Sasao et al. (1980) and Wahr and Bergen (1986) provided the theoretical estimation values of FCN parameters. Gwinn, Herring and Shapiro (1987) first obtained the observational values of FCN parameters by very long base Interference (VLBI) at Cambridge University. In the same year, Neuberg and Zürn in former West Germany and Hinderer in France began to retrieve FCN parameters by the observation of gravity tides and introduced the stacking method. The other scholars who researched into the same geophysical problems by applying the data of gravity tides basically followed the stacking method. The results they reached were similar to the observational result of FCN parameters given by Neuberg et al. in 1987. But the observational results of FCN parameters gained from gravity tides were not identical with those from VLBI, mainly because of the large difference of quality of FCN. So there was not an affirmative observational result of FCN parameters since then. In this paper, The authors firstly introduce the tri-frequency spectrum method with clearly geometrical and geophysical meaning for the resolution of FCN parameters, and the observational results of FCN parameters obtained from tide data at three superconducting gravity stations were accordant with those from VLBI, which will be relatively important to arriving at a certain observational result of FCN parameters.     

Tri-frequency spectrum method and results for resolving the parameters of Earth's liquid core free nutation

The parameters of Earth free core nutation (FCN) are two relatively significant geophysical parameters. Sasao et al. (1980) and Wahr and Bergen (1986) provided the theoretical estimation values of FCN parameters. Gwinn, Herring and Shapiro (1987) first obtained the observational values of FCN parameters by very long base Interference (VLBI) at Cambridge University. In the same year, Neuberg and Zürn in former West Germany and Hinderer in France began to retrieve FCN parameters by the observation of gravity tides and introduced the stacking method. The other scholars who researched into the same geophysical problems by applying the data of gravity tides basically followed the stacking method. The results they reached were similar to the observational result of FCN parameters given by Neuberg et al. in 1987. But the observational results of FCN parameters gained from gravity tides were not identical with those from VLBI, mainly because of the large difference of quality of FCN. So there was not an affirmative observational result of FCN parameters since then. In this paper, The authors firstly introduce the tri-frequency spectrum method with clearly geometrical and geophysical meaning for the resolution of FCN parameters, and the observational results of FCN parameters obtained from tide data at three superconducting gravity stations were accordant with those from VLBI, which will be relatively important to arriving at a certain observational result of FCN parameters.     

Indoor scene texturing based on single mobile phone images and 3D model fusion

Realistic texture mapping and coherent up-to-date rendering is one of the most important issues in indoor 3-D modelling. However, existing texturing approaches are usually performed manually during the modelling process, and cannot accommodate changes in indoor environments occurring after the model was created, resulting in out-dated and misleading texture rendering. In this study, a structured learning-based texture mapping method is proposed for automatic mapping a single still photo from a mobile phone onto an already-constructed indoor 3-D model. The up-to-date texture is captured using a smart phone, and the indoor structural layout is extracted by incorporating per-pixel segmentation in the FCN algorithm and the line constraints into a structured learning algorithm. This enables real-time texture mapping according to parts of the model, based on the structural layout. Furthermore, the rough camera pose is estimated by pedestrian dead reckoning (PDR) and map information to determine where to map the texture. The experimental results presented in this paper demonstrate that our approach can achieve accurate fusion of 3-D triangular meshes with 2-D single images, achieving low-cost and automatic indoor texture updating. Based on this fusion approach, users can have a better experience in virtual indoor3-D applications.     

Investigation of the time variability of diurnal tides and resonant FCN period

The time variability of diurnal tides was investigated by analyzing gravity observations from global superconducting gravimeter (SG) stations with running time intervals. Through least-square and Bayesian approaches, FCN resonance parameters were estimated for each data section after obtaining the tidal parameters of mainly diurnal tidal waves. The correlation of the time variation in diurnal tidal waves and FCN period was discussed. For comparison, a similar method was used to analyze VLBI observations to study the time variability of nutation terms and FCN period. The variation trend of the FCN period totally depends on the Ψ1 wave in tidal gravity and on the retrograde annual term in nutation. We observed a similar variation trend in the FCN periods obtained from different SG stations worldwide and VLBI observations. The relation between diurnal tides and LOD variations is discussed and the possible mechanisms of the decadal variation in FCN periods were discussed.     

面向多源数据地物提取的遥感知识感知与多尺度特征融合网络

房屋等建筑物的分布情况是衡量地区发展的重要指标,利用遥感影像实现建筑物的自动高精度提取在指导城乡规划和市镇建设等方面具有重要意义。已有方法大多忽略了像素数较少的小面积建筑和边缘等细节信息的处理,针对此问题,提出了一种双路细节关注网络,将语义特征与细节关注特征双路并行优化,进一步提高了遥感影像中建筑物的提取精度。所提方法首先使用双路特征提取模块获取语义特征与细节关注特征,并在解码过程中进行双向优化,增强语义特征细节的同时提高细节关注特征的连续性与类别准确性,然后对二者进行融合,结合细节关注损失的监督,实现建筑物的高精度提取。在WHU建筑物数据集、ISPRS Vaihingen数据集与某地区国产高分数据集上,将所提方法与多种主流方法进行了对比验证,所提方法的F1分数和交并比均高于对比方法,且提取的建筑物完整性更好,小面积建筑漏检、误检率更低。

     

多尺度SLIC-GMRF与FCNSVM联合的高分影像建筑物提取

遥感影像建筑物提取具有重要的应用价值。然而,高分辨率遥感影像中细节信息繁多、特征复杂,增加了建筑物提取难度。针对这一问题,本文提出一种基于多尺度SLIC-GMRF和FCNSVM的建筑物提取方法,一定程度上提高了高分辨率遥感影像建筑物提取能力。首先,利用多尺度SLIC-GMRF分割算法确定初始建筑物区域,然后,充分利用FCN神经网络在语义分割中的优势抽取建筑物特征,最后,结合提取出的建筑物特征训练SVM分类器细化建筑物提取结果,通过3种控制实验,两种对比方法得出以下结论:SLIC分割算法影响初始分割结果;SVM分类器影响建筑物细部提取;FCN特征影响SVM分类器性能。对于特征清晰、遮挡干扰较少的研究区,本文方法能够较好提取影像中的建筑物,查准率、查全率、质量指标均优于对比方法,对建筑物复杂分布的研究区同样能够取得较好的提取效果。