单帧图像的超分辨率技术进展

单帧图像超分辨率重建是就是由一幅降质的含噪的低分辨率图像来估计一幅高分辨率图像。列举了当前研究的几种主要算法,从峰值信噪比、框架相似性和边缘稳健性的标准来加以比较。最后的结果显示RS算法性能最优,这也符合主观评价标准。     

基于迁移学习的原子力显微镜成像恢复方法

受限于探针针尖结构尺寸,用原子力显微镜进行微纳测量时会产生图像边缘失真.提出了一种基于迁移学习的原子力显微镜成像恢复方法,通过迁移学习训练源模型和靶模型实现一维栅格成像恢复.该方法采用数学形态法中的腐蚀算法生成栅格点云数据,通过U-Net网络源模型从点云中提取针尖卷积效应的特征向量,将权重参数迁移至U-Net网络靶模型,靶模型在自适应正则化方法下进行监督学习.实验结果表明,该方法能有效恢复一维栅格的原子力显微镜测量图像,提高横向分辨力,可用于纳米栅格的线宽检测上.     

1980—2018年镇赉县湿地景观格局变化与人为干扰的关系

以吉林省最大湿地保留区镇赉县为研究区，基于1980—2018年8期土地利用数据，采用网格分析法研究湿地景观格局变化规律及景观指数与人为干扰度的关系。结果表明：（1）镇赉县湿地面积减少290.7 km2，水体、滩地、沼泽变化率分别为-0.45，-2.97和-4.84 km²·a^-1。（2）镇赉县湿地率（PLAND）下降6.1%，且从西北部向东南部逐渐递增；斑块数量（NP）与斑块平均面积（AREA_MN）均呈先上升后下降的趋势，东南部湿地景观破碎化加剧，中部破碎化减缓；湿地连通度的空间分布格局为西北向东南递增，斑块结合度指数（COHESION）在中部地区显著增加，在东南部地区缓慢降低。（3）镇赉县湿地人为干扰度逐渐增强、耕地面积的增加和交通网的扩张是镇赉县湿地面积萎缩、景观破碎化加剧和连通度降低的主要人为原因。人为干扰度与NP呈显著正相关，与PLAND、AREA_MN、COHESION均呈负相关。稳定西南部和东北部耕地中湿地斑块，进一步加强中部湿地水体连通性，排查东南部湿地核心区违规建筑，实施分区治理，加强对镇赉县现有湿地的保护，以期实现区域可持续发展。     

Sphaerococcus coronopifolius (Rhodophyta,Gigartinales): a Mediterranean red alga with potential and applications in restoration

Abstract

Experimental studies conducted on some species of Mediterranean red algae allowed to identify Sphaerococcus coronopifolius Stackhouse as a valid alternative to the Pacific alga Gloiopeltis furcata (Postels & Ruprecht) J. Agardh, for the extraction of a material usable as natural consolidant and adhesive in the field of restoration. Promising results have been observed by comparing the extracts obtained from these two algae after the same extraction procedure. Chemical analysis (FTIR) revealed that S. coronopifolius has qualities similar to G. furcata. Even more promising results for S. coronopifolius compared to G. furcata were observed after the analysis of pH and conductivity, and the adhesion tests carried out on both extracts.     

硬X射线焦斑环孔编码成像的理论与实验研究

针对几十~百keV能量的低强度X射线焦斑源测量,建立了高探测效率环孔编码成像技术。研究给出了环孔成像效率和信噪比与环孔结构参数的关系,采用微联结的环孔结构设计并模拟了联结区尺寸对成像质量的影响,解决了环孔结构同轴成像技术难题。所建立环孔成像系统应用于Unique-II X射线焦斑源成像实验。结果表明,X射线焦斑的空间分布近似为‘马鞍’形状,而且中间部分的强度低于两侧的强度,与针孔成像结果相似,但环孔的探测效率明显高于针孔的效率。最后,分析了维纳滤波与R-L两种复原图像方法的效果。     

改进的PNJL模型下QCD的相图

Polyakov-Nambu-Jona-Lasinio（PNJL）模型是研究强相互作用物质性质的使用最为广泛的有效模型之一。在PNJL模型的基础上考虑了手征凝聚和Polyakov圈之间的纠缠作用,并且引入了化学势修正的Polyakov有效势,由此得到了化学势依赖的entangled PNJL（μEPNJL）模型。在平均场框架下的计算结果表明:相较于原始的PNJL模型,由μEPNJL模型计算得到的临界点（CEP）朝着温度更高、化学势更小处移动,并且手征对称性恢复相变和退禁闭相变在较大的化学势范围内都重合得很好。通过与STAR合作组在相对论重离子对撞机（RHIC）上进行的净质子数分布的测量结果相比,可以发现,通过适当的参数调节,由μEPNJL模型计算得到的CEP更加靠近实验预言的CEP可能存在的区域。Polyakov-Nambu-Jona-Lasinio (PNJL) model is one of the most popular effective quark models to investigate the properties of strongly interacting matter. Based on the PNJL model, we consider the entanglement interactions between the chiral condensate and Polyakov-loop, as well as the chemical potential modification of Polyakov-loop potential simultaneously, which is named μEPNJL model. Compared with the original PNJL model, the calculations in the mean field approximation show that the critical end point (CEP) given in the μEPNJL model moves towards higher temperature and smaller chemical potential in the T-μ phase diagram. Besides, the chiral symmetry restoration and deconfinement phase transition coincide well in a wide range of chemical potential. Comparing our calculations with the measurement of the moments of net-proton multiplicity distributions at Relativistic Heavy-Ion Collider (RHIC) by STAR Collaboration, we find that the CEP given by μEPNJL model can be closer to the range predicted by the experiment through appropriate parameter adjustment.     

一种基于星座恢复的相干光OFDM系统非线性抑制方法

鉴于相干光正交频分复用系统（CO-OFDM）易受非线性效应影响,严重制约系统传输性能, 提出一种基于星座恢复的降低系统峰均值功率比（PAPR）的方法。对系统模型及作用机理进行分析,仿真结果显示,以10 Gbit/s的速率通过单模光纤传输720 km,基于星座恢复的CO-OFDM系统Q值较无非线性抑制的CO-OFDM系统有3 dB提高;另外,系统最佳限幅比例系数受色度色散影响,随着色散因子的增大,星座恢复效果也逐渐减弱,在传输240 km条件下,色散因子为12 ps/nmkm较6 ps/nmkm,限幅比例系数从0.9降低至0.8。     

基于Poisson-Markov分布最大后验概率的多通道超分辨率盲复原算法

在Poisson和Markov随机场分布假设基础上,改进了存在噪声时多通道互限制原则,提出了一种基于最大后验概率判据的多通道图像盲复原算法,可在无需已知通道内离散点扩散函数性质、类型和具体分布等先验知识的情况下,在图像幅度限制约束空间和离散点扩散函数能量恒定约束空间内,通过对图像和离散点扩散函数的交替迭代,使两者收敛到全局最优解,最终复原出超分辨率图像.对计算机模拟和实际存在微平移、微离焦情况的图像处理表明：算法在不同信噪比水平和非命中PSF尺寸下具有良好的图像复原效果和稳定性. 关键词： 泊松-马尔科夫分布 最大后验概率 多通道 盲复原     

Raman and infrared analysis of glues used for pottery conservation treatments

The conservation of ceramics and glass involves the repair or the restoration of broken and fragmented artefacts with polymer adhesives and gap fillers. In the past, many different adhesives had been used. Re‐restoration is often required and the fast identification of adhesive residues on objects would be very useful to define the best of way to remove them, in particular to avoid the use of noxious chemicals. Sixteen pottery artefacts restored during the 19th and 20th centuries at the Musée National de Céramique in Sèvres have been analysed by non‐destructive Raman microspectroscopy. For comparison purposes, the artefacts were also sampled in order to acquire infrared (IR) absorption spectra in KBr pellets. Modern adhesives (methyl metacrylate/acrylate, vinyl acetate, polychloroprene, methyl 2‐cyanoacrylate and diglycidylether biphenol) were also characterised with the same methods. IR and Raman spectra were obtained for all ancient glue residues, but among the 16 analysed items, only 7 adhesives have been identified unambiguously, and an assignment proposed for 4 others solely on the basis of the Raman signature. Copyright © 2009 John Wiley & Sons, Ltd.     

基于约束共轭梯度的高能闪光照相图像复原算法

 针对闪光照相系统模糊较大、成像信噪比低的特点,提出了一种基于约束共轭梯度的闪光照相图像复原算法,将闪光照相图像复原问题转化为一个约束优化问题,引入基于非负、中值滤波和偏微分方程的光滑约束条件,并利用约束共轭梯度法迭代求最优解。数值试验表明,该算法能较好再现图像边缘信息,复原出的图像在信噪比和视觉方面都有较大提高。