基于全卷积网络的高分辨遥感影像目标检测   总被引：2，自引：0，他引：2  

徐逸之  姚晓婧  李祥  周楠  胡媛 《测绘通报》2018,(1):77-82

目标检测是遥感图像分析处理中的研究热点之一,具有十分重要的科研和应用价值。传统遥感影像目标检测方法多使用人工构造的浅层次特征,结合支持向量机、随机森林、Adaboost等分类器进行目标识别,难以充分挖掘和利用影像中的深层特征。近年来,深度学习,特别是卷积神经网络在图像认知方面取得了巨大成功。在目标检测领域,以Faster R-CNN算法为代表的方法取得了突破性进展,检测精度大幅提高,检测速度达到了近实时的性能。但是,Faster R-CNN算法由于使用了感兴趣区域（RoI）池化层,各个RoI计算不共享,因此检测速度依然有待提高。R-FCN基于全卷积网络结构,同时采用位置敏感池化来引入平移变化,抵消全卷积网络造成的平移不变形问题,检测精度和效率都有了很大的提高。本文阐述了R-FCN算法原理,并运用于高分辨遥感影像目标检测分析了不同参数和网络结构对R-FCN检测效果的影响,比较了利用Fast R-CNN、Faster R-CNN和R-FCN 3种算法进行飞机识别的性能。试验结果表明,利用R-FCN进行飞机识别定位可以达到99.3%的准确率和每张图180 ms的检测速度。  相似文献   

论地球系统四维成矿及矿床学研究趋向 ——七论矿床的成矿系列     

陈毓川  裴荣富  王登红  黄凡 《矿床地质》2020,39(5):745-753

矿床的成矿系列（简称“成矿系列”）概念是地球系统四维成矿的学术思想。成矿系列亦是矿床的一种新的分类——矿床自然分类。成矿系列概念的创新点：突出成矿在特定的时间域和特定的空间域（地质构造环境）,作为四维时空域整体进行成矿作用的成矿过程及其形成矿床的研究;厘定地球系统演化各个时空域中的成矿实体——矿床成矿系列,给每个矿床在地球演化的时空域中定位;开展各时空域成矿体之间的内在联系、组成的各层次成矿体系及地球成矿体系的研究;确定5类成矿作用（岩浆、变质、沉积、表生及非岩浆非变质含矿流体成矿作用）;建立成矿系列的5个序次,同时构成新的矿床自然分类体系。提出矿床学研究趋势与内容：深化矿床成因及其形成过程研究、地球系统四维时空域中区域成矿研究、急需战略性矿产研究、矿产预测、勘查与发展矿业的研究、配合找矿技术方法研发与应用。  相似文献   

一种基于卷积运算的遥感影像道路提取方法     

杨先武  韦春桃  吕健刚 《地理空间信息》2010,8(4):87-89

根据遥感影像道路特征,提出一种基于卷积运算和数学形态学的道路提取方法。首先通过拉普拉斯算子对图像行锐化处理;再用卷积运算获得线特征加强系数,在加强道路线特征的同时,背景灰度恒定区域被弱化;然后去除短线和块状噪声,适当地加入人工干预进行道路修剪;最后用数学形态学细化处理得到道路网。实验结果表明,该方法能有地从遥感影像中提取出道路。  相似文献   

转换波Kirchhoff叠前时间偏移的成像优化方案          下载免费PDF全文

岳玉波  李振春  钱忠平  孙鹏远  杨雪霖  薛贵仁 《地球物理学报》2018,61(3):1188-1195

叠前时间偏移是转换波数据处理流程中的重要技术环节.由于下行P波和上行S波对应着不同类型的传播速度和传播机理,使得转换波偏移的诸多实施环节和实现技巧均有别于常规的纵波偏移.本文就VTI介质转换波Kirchhoff叠前时间偏移实现过程中的几个关键技术环节进行研究分析,并给出了相应的实施方案.首先,依据常速假设推导了简化的转换波时间偏移振幅加权函数,在保证偏移的高效计算的同时,兼顾了成像振幅的准确性;其次,依据直射线近似假设,推导了转换波偏移的最大无假频频率计算公式,确保了算子假频的有效滤除;接下来,设计使用了依据共转换点对称的偏移孔径,保证了偏移孔径的优化选取.此外,本文还提出利用三次卷积插值进行成像网格上参数值的插值运算,以避免传统线性插值算法的非平滑性对偏移剖面中高频信息的不良影响.多个算例的应用效果证明,本文所给出的关键技术环节的实施方案可以有效的提升转换波叠前时间偏移的成像效果.  相似文献   

Seismic modelling for reservoir studies: a comparison between convolutional and full-waveform methods for a deep-water turbidite sandstone reservoir     

Hamed Amini  Colin MacBeth  Asghar Shams 《Geophysical Prospecting》2020,68(5):1540-1553

Two seismic modelling approaches, that is, two-dimensional pre-stack elastic finite-difference and one-dimensional convolution methods, are compared in a modelling exercise over the fluid-flow simulation model of a producing deep-water turbidite sandstone reservoir in the West of Shetland Basin. If the appropriate parameterization for one-dimensional convolution is used, the differences in three-dimensional and four-dimensional seismic responses from the two methods are negligible. The key parameters to ensure an accurate seismic response are a representative wavelet, the distribution of common-depth points and their associated angles of incidence. Conventional seismic images generated by the one-dimensional convolutional model suffer from lack of continuity because it only accounts for vertical resolution. After application of a lateral resolution function, the convolutional and finite-difference seismic images are very similar. Although transmission effects, internal multiples and P-to-S conversions are not included in our convolutional modelling, the subtle differences between images from the two methods indicates that such effects are of secondary nature in our study. A quantitative comparison of the (normalized root-mean-square) amplitude attributes and waveform kinematics indicates that the finite-difference approach does not offer any tangible benefit in our target-oriented seismic modelling case study, and the potential errors from one-dimensional convolution modelling are comparatively much smaller than the production-induced time-lapse changes.  相似文献   

Application on Anomaly Detection of Geoelectric Field Based on Deep Learning     

WEI Lei  AN Zhanghui  FAN Yingying  CHEN Quan  YUAN Lihu  HOU Zeyu 《中国地震研究》2020,34(3):358-377

The deep learning method has made nurnerials achievements regarding anomaly detection in the field of time series. We introduce the speech production model in the field of artificial intelligence, changing the convolution layer of the general convolution neural network to the residual element structure by adding identity mapping, and expanding the receptive domain of the model by using the dilated causal convolution. Based on the dilated causal convolution network and the method of log probability density function, the anomalous events are detected according to the anomaly scores. The validity of the method is verified by the simulation data, which is applied to the actual observed data on the observation staion of Pingliang geoeletric field in Gansu Province. The results show that one month before the Wenchuan M_S8.0, Lushan M_S7.0 and Minxian-Zhangxian M_S6.6 earthquakes, the daily cumulative error of log probability density of the predicted results in Pingliang Station suddenly decreases, which is consistent with the actual earthquake anomalies in a certain time range. After analyzing the combined factors including the spatial electromagnetic environment and the variation of micro fissures before the earthquake, we explain the possible causes of the anomalies in the geoelectric field of before the earthquake. The successful application of deep learning in observed data of the geoelectric field may behefit for improving the ultilization rate both the data and the efficiency of detection. Besides, it may provide technical support for more seismic research.  相似文献   

基于双流残差卷积神经网络的养殖鳗鲡(Anguilla)摄食强度评估研究     

李凯  江兴龙  许志扬  林茜 《海洋与湖沼》2023,54(4):1207-1216

为实现对养殖鳗鲡(Anguilla)摄食强度的准确评估,提出了一种基于双流残差卷积神经网络的鳗鲡摄食强度评估方法,该方法针对传统双流网络(Two-stream)中存在的问题做出了相应的改进。首先针对传统双流网络存在网络结构较浅,无法提取到充分的鳗鲡摄食行为特征的问题,选择使用ResNet50网络进行替换,以提取到更具代表性的特征。其次针对传统双流网络最后的分类结果是把空间流和时间流的得分取平均值融合而获得,这种方式较为简单,且其空间流和时间流网络为独立进行训练,容易导致网络出现学习不到鳗鲡摄食行为的时空关联特征的问题,选择使用特征层融合方式对空间流和时间流网络提取获得的特征进行融合,让网络能够并行进行训练,以提取到时空信息间的关联特征。试验结果表明:文内提出的基于双流残差卷积神经网络的鳗鲡摄食强度评估方法准确率达到98.6%,与单通道的空间流和时间流网络相比,准确率分别提升了5.8%和8.5%,与传统的双流网络相比准确率也提升了3.2%。  相似文献   

雷达径向风资料的四维变分同化试验   总被引：8，自引：2，他引：6  

张林  倪允琪 《大气科学》2006,30(3):433-440

在MM5四维变分同化系统(MM5 4D-Var)的基础上开发出雷达径向风资料的同化模块,利用改进后的同化系统对2002年7月22日湖北宜昌多普勒雷达的径向风观测资料进行四维变分同化试验,分析雷达径向风资料的同化对中尺度数值模式初始场的调整以及对中尺度强降水模拟的影响.研究结果表明,同化雷达径向风资料加强了中尺度对流系统低层100km左右范围内的风场辐合.调整后的风场具有更明显的β中尺度特征.利用14 min的雷达径向风资料可以改进3 h之内的强降水模拟,尤其是对雷达站东南侧强降水的模拟.  相似文献   

基于改进型U-Net网络的高分辨率遥感影像建筑物提取     

吕道双  林娜  冯丽蓉  张小青 《地理空间信息》2021,19(1):30-34

针对传统人工提取方法自动化程度低、过分依赖人工设计的特征,以及现有的深度学习方法中存在的提取精度不高等问题,提出了一种基于改进型U-Net网络的高分辨率遥感影像建筑物提取方法.首先将空洞卷积加入到网络中,利用不同尺度的空洞卷积对来自网络编码部分的结果进行多尺度特征提取;再对提取的特征进行特征融合,并输入到网络的下一层;...  相似文献   

Theoretical link between rainfall and flood magnitude          下载免费PDF全文

Robert E. Criss 《水文研究》2018,32(11):1607-1615

The rainfall–run‐off convolution integral is analytically solved for several models for the elementary hydrograph. These solutions can be combined with available rainfall frequency analyses to predict flood flows along streams for different recurrence intervals, using no free parameters for gauged streams and one estimable parameter for ungauged streams. Extreme discharge magnitudes at gauged sites can be typically estimated within a factor of two of actual records, using no historical data on extreme flows. The flow predictions reproduce several important characteristics of the flood phenomenon, such as the slope of the regression line between observed extreme flows and basin area on the conventional logQ versus logA plot. Importantly, for the models and data sets investigated, the storm duration of greatest significance to flooding was found to approximate the intrinsic transport timescale of the particular watershed, which increases with basin size. Thus, storms that deliver extraordinary amounts of rainfall over a particular time interval will most greatly activate basins whose time constants approximately equal that interval. This theoretical finding is supported by examination of the regional hydrological response to the massive storms of September 14, 2008, and April 28–30, 2017, which caused extraordinary record flooding of basins of about 5–100 km² and 500–4,000 km², respectively, but produced few records in basins that were larger or smaller than those ranges.  相似文献