A-FPN算法及其在遥感图像船舶检测中的应用

光学遥感图像船舶检测主要面临两个挑战:光学遥感图像背景复杂,船舶检测易受海浪、云雾及陆地建筑等多方面干扰;遥感图像分辨率低,船舶目标小,对于其分类与定位带来很大困难;针对上述问题,在FPN的基础上,提出一种融入显著性特征的卷积神经网络模型A-FPN (Attention-Based Feature Pyramid Networks)。首先,利用卷积提取图像特征金字塔;然后,利用顶层金字塔逐级构建显著特征层,抑制背景信息,通过金字塔顶层的细粒度特征提高浅层特征的表达能力,构建自上而下的多级显著特征映射结构;最后利用Softmax分类器进行多层级船舶检测。A-FPN模型利用显著性机制引导不同感受下的特征进行融合,提高了模型的分辨能力,对遥感图像处理领域具有重要应用价值。实验阶段,利用公开的遥感目标检测数据集NWPU VHR-10中的船舶样本进行测试,准确率为92.8%,表明A-FPN模型适用于遥感图像船舶检测。     

On the nature of the Galactic nebula We 1–12

In 1977, Weinberger published a list of 12 new extended possible planetary nebulae (PNe). Whereas, because of their characteristic morphology and/or the presence of a blue central star, almost all of them could easily be suspected to be genuine planetary nebulae, one object (No. 12) captivated because of its unusually bright central star. This find prompted Kaler & Feibelman to question the PN nature of this object (We 1–12) on the basis of IUE spectra. A definite conclusion could, however, not be drawn by them; thus, until now, the real nature of We 1–12 remained unsolved. For the first time, a spectral investigation of both the central star and the nebula is presented in this paper. It definitely shows that this intriguing object is an (isolated) H  ii region and its central star, as previously assumed, is an early B star which serves as the ionizing source. We 1–12, a part of which is coincident with a weak IRAS point source, is at a distance of 2–2.6 kpc and is reddened by E ( B − V )=0.6–0.8 mag.     

精密单点定位估计GPS卫星的P1-C1码偏差及稳定性分析

给出了利用精密单点定位（PPP）技术估计GPS卫星P1-C1码偏差的数学模型,并以BRUS、GODE、SHAO和NIST四个跟踪站2010年10月份一个月的观测数据为例,采用PPP方法计算了所有GPS卫星的P1-C1码偏差,并与欧洲定轨中心提供的P1-C1码偏差进行了比较,结果表明：四个站估计的P1-C1码偏差精度均可达到几个厘米。一个月的计算结果表明：卫星的P1-C1码偏差在一个月内变化平缓。     

基于Sentinel-1的典型地物时间失相干分析

为了探究影响Sentinel-1时间失相干的因素,实验选取美国南加州柯汶纳市及其周边区域进行研究,使用50景Senti-nel-1数据,分析了不同地物的时间失相干现象,并研究了风雨等天气因素及不同极化方式对相干性的影响.主要得出以下结论:Sentinel-1数据典型地物的失相干特征区别较大,可以较好地指导地物识别、辅助...     

近海水体悬浮物HJ-1 号小卫星CCD 定量反演研究

基于H J-1号小卫星CCD数据,开展近海水体悬浮物含量监测研究.采用邻近清洁水体和同日MODIS气溶胶产品的方法对CCD辐亮度数据进行较精准的大气校正;利用得到的水体遥感反射率,结合地面准同步实测悬浮物含量数据建立悬浮物反演模型,获得研究区悬浮物的空间分布.模型的相关系数R2为0.849,平均误差为33.0％,反演结果较为理想.结果表明,HJ-1号小卫星作为中国首个灾害监测小卫星星座,能够实现定量反演近海水体的悬浮物含量,对中国近海水体水质的监测和治理具有重要意义.     

Flood‐based analysis of high‐magnitude sediment transport using a non‐parametric method

Upland erosion and the resulting reservoir siltation is a serious issue in the Isábena catchment (445 km² Central Spanish Pyrenees). During a three‐month period, water and sediment fluxes have been monitored at the catchment outlet (Capella), two adjacent subcatchments (Villacarli, 41 km²; Cabecera, 145 km²) and the elementary badland catchment Torrelaribera (8 ha). This paper presents the results of the monitoring, a method for the calculation of a sedigraph from intermittent measurements and the derived sediment yields at the monitored locations. The observed suspended sediment concentrations (SSCs) demonstrate the role of badlands as sediment sources: SSCs of up to 280 g l^?1 were encountered for Villacarli, which includes large badland areas. SSCs at the Cabecera catchment, with great areas of woodland, barely exceeded 30 g l^?1. SSCs directly at the sediment source (Torrelaribera) were comparable to those at Villacarli, suggesting a close connection within this subcatchment. At Capella, SSCs of up to 99 g l^?1 were observed. For all sites, SSC displayed only a loose correlation with discharge, inhibiting the application of a simple sediment rating curve. Instead, ancillary variables acting as driving forces or proxies for the processes (rainfall energy, cumulative discharge, rising/falling limb data) were included in a quantile regression forest model to explain the variability in SSC. The variables with most predictive power vary between the sites, suggesting the predominance of different processes. The subsequent flood‐based calculation of sediment yields attests high specific sediment yields for Torrelaribera and Villacarli (6277 and 1971 t km^?2) and medium to high yields for Cabecera and Capella (139 and 410 t km^?2) during the observation period. In all catchments, most of the sediment was exported during intense storms of late summer. Later flood events yield successively less sediment. Relating upland sediment production to yield at the outlet suggests considerable effects of sediment storage within the river channel. Copyright © 2008 John Wiley & Sons, Ltd.