分数阶傅里叶变换的快速计算新方法

本文提出了一种分数阶傅里叶变换(FRFT)高分辨(Zoom-FRFT)算法,通过设置谱区间和输出点数M,可实现任意局部谱的高分辨计算.随后,针对M很小时Zoom-FRFT运算效率低的问题,提出了基于Horner的单点快速计算(SP-FRFT)方法,并针对零点计算做出进一步简化.利用SP-FRFT可提高少量点输出时的计算效率,也可用于非均匀采样点计算.仿真结果验证了算法的有效性.     

基于改进U-Net网络的腺体细胞图像分割算法

针对腺体图像在自动分割过程中由于多尺度目标和信息丢失影响导致准确率降低的问题,文中采用了一种引入注意力模块的全卷积神经网络模型。该模型遵循编码器-解码器结构,在编码网络中用空洞残差卷积层代替原有的普通卷积层,并添加空洞金字塔池;再在解码网络中加入注意力模块,使模型输出高分辨率特征图,提高对多尺度目标的分割精度。实验结果表明,提出的网络模型参数少分割精度高,对腺体图像的平均分割精度高达89.7%,具有较好的鲁棒性。     

无卷边单分子层WS2纳米片原子结构的高分辨电子显微术研究

本文利用高分辨电子显微术(HRTEM)确定了无卷边的单分子层WS2纳米片的原子结构像.通过对样品不同区域HRTEM像进行傅里叶变换(FFT)分析,得到了各个区域的WS2纳米片的晶体取向,确定了WS2纳米片分子层数分布.根据电子显微像强度与样品层数的线性关系以及WS2纳米片的单分子层数分布模型,确定不同区域WS2纳米片的分子层数.为了进一步定量分析结果的正确性,通过模拟电子显微像在不同碳膜厚度及不同成像条件下WS2纳米片的HRTEM像,也确定了在像衬度上与实验像很好的匹配.     

调频连续波SAR的研究进展

调频连续波合成孔径雷达(SAR)是一种新体制的高分辨率成像雷达,具有体积小,重量轻,耗电少,造价低,分辨率高等优点,非常适合小型化SAR的应用。文中建立了更易于实现的调频连续波SAR回波信号模型,给出了基本的成像处理流程;综述了国内外调频连续波SAR的最新进展,讨论了可能的应用背景,分析了进一步发展面临的主要技术问题。分析表明,调频连续波SAR具有广阔的应用前景。     

用于超宽带高分辨率成像的二维稀疏MIMO面阵拓扑结构研究

分析并总结了超宽带二维多输入多输出(Multiple Input Multiple Output, MIMO)面阵拓扑结构设计的两条原则——等效孔径的均匀性与无明显遮蔽性, 并根据这两条原则提出了一种用于超宽带近距离高分辨率成像的新型面阵拓扑结构.与尺寸、阵元数相同的MIMO面阵相比, 该新型面阵结构在仿真获取的方向图中具有更好的聚焦效果和旁瓣抑制能力.并且, 不同距离下的聚焦结果显示, 该面阵的峰值旁瓣水平均要低于另两个阵列2 dB以上.对复杂目标成像的实验结果进一步证明了该阵列良好的成像性能.结合其等效阵元数量较少的特点, 文中提出的这种新型MIMO面阵拓扑结构为高效、实时的超宽带近距离高分辨率成像应用提供了可能.     

多方位高分辨率SAR的三维目标自动重建(一)目标像的检测

用多方位SAR图像融合获取立体目标信息是高分辨率SAR的一项关键应用技术.以t长方体为建筑物目标模型,提出从多方位米级分辨率SAR图像自动重建三维立体建筑物目标的方法.全文分两部分,第一部分介绍从米级分辨率SAR图像检测提取建筑物目标像的方法.先用恒虚警率(CFAR)检测器检测各方位SAR图像中建筑物目标像的边缘,然后用脊滤波细化边缘,再根据条状建筑物像的直线特征,用分块平行线Hough变换检测平行线段,最终检测得到平行四边形的建筑物目标像.文中用虚拟场景模拟的SAR图像和机载全极化X波段Pi-SAR图像做试验,提取了建筑物目标像.第二部分以本文为基础进行三维建筑物目标群的重建.     

宽带随机混沌雷达高速目标速度估计与成像

随机混沌具有真随机性、对初值敏感、易于产生和控制等特点,频率步进信号易于工程实现和处理,结合两者的优势,提出了一种载频随机步进的随机混沌信号(RSCFSCS)模型,用于高速目标的速度估计和距离维高分辨成像。首先,通过非周期函数激励非线性系统,产生不可预测的随机混沌信号(SCS),经频率调制后用作基带子脉冲。同时,将SCS通过映射变换得到跳频编码(FHC),用来决定调频脉冲串的载频步进。RSCFSCS 速度估计包括粗搜索和精搜索,粗搜索采用固定步长,保证速度偏差小于速度分辨单元,而精搜索采用黄金分割搜索算法可得到精确的速度估计。最后,子脉冲经相干合成形成宽带信号,实现高分辨距离成像。数值仿真表明提出的信号模型和处理算法性能良好。     

Development of an HRTEM image analysis method to quantify carbon nanostructure

An analytical method for analyzing high-resolution transmission electron microscopy (HRTEM) images is presented. The method is composed of two major parts: digital image processing and lattice fringe characterization. The digital image processing is composed of the following operations: negative transformation, region of interest (ROI) selection, contrast enhancement, Gaussian lowpass filter, top-hat transformation (used to correct uneven illumination across an image), thresholding to obtain a binary image, morphological modification, clearing fringes on the ROI border, skeletonization, and removing short fringes that lack physical meaning. These operations are defined by equations with effects illustrated by images. The lattice fringe characterization generates statistics on fringe length, tortuosity, and separation based on the skeletons of the graphene layers. Fringe length and tortuosity are obtained automatically from the features of the skeletons, while fringe separation is obtained by analyzing manually selected fringe pairs. The algorithms are implemented within MATLAB, and demonstrated upon an exemplary HRTEM image of carbonaceous material. The image analysis algorithm permits quantitative HRTEM, here specifically addressing nanostructure of carbonaceous materials.     

In vivo lipolysis induced by heparin injection or fasting: a proton MR spectroscopy analysis of human plasma

The impact of induced lipolysis on the composition of plasma lipids is analyzed by proton magnetic resonance spectroscopy (MRS) in humans. The variations of the methylene and methyl resonances from lipids in lipoproteins are studied under two sets of lipolytic conditions: acute endovascular lipolysis induced by an intravenous injection of heparin and subacute lipolysis induced by short fasting. During acute lipolysis, the degradation of the very low density lipoproteins structures is well correlated to the modifications observed in the areas of CH₂ and CH₃ MRS signals. The comparison of regular spectra, spectra with water signal suppression, and spectra recorded with a spin-echo sequence provides information on the behavior of the different parts of the lipoproteins, that is, the neutral core, little affected by heparine-induced lipoprotein lipase (LPL) activation, and the surface layer supplying substrates to LPL. During 48 h of fasting, only limited modifications occur on the MR spectra, and lipolysis cannot be documented in details.Address for correspondence: CRMBM-CNRS, Faculté de Médecine, 27, bd Jean Moulin, 13005 Marseille, France.     

MR high-resolution imaging of arterial intimal hyperplasia at 1.5 T

Intimal hyperplasia is the earliest microscopic change detectable in the arterial wall in the development of arteriosclerosis and atherosclerosis. To enable high-resolution imaging of this, a standard 1.5-T machine was modified using inserted gradient coils of 30 cm internal diameter and surface radio-frequency coils. Six rabbit aorta specimens with injury-induced intimal hyperplasia and two normal rabbit aorta specimens were imaged using a spin-echo sequence of TE 40. Pixel sizes as small as 20 µ×20 µ could be obtained and the area of intimal hyperplasia could be measured at 80 µ×40 µ. The results were compared with those made by a computer-linked microscope. In the injured aortas, there was a low-signal region on magnetic resonance imaging which corresponded to the media of the vessel wall when compared with the histology.T ₁ andT ₂ constants for the media were determined and compared with those of a normal specimen. TheT ₁ constant was shorter for the media of the injured aorta than in the normal. This could be related to the disruption of the internal elastic lamina and migration of muscle cells to the intima which occurs following injury. The popliteal artery of four normal volunteers was imaged and the area of the wall could be measured at a pixel size of 312 µ×160 µ. This suggests thatin vivo studies of intimal hyperplasia arteriosclerosis and atherosclerosis could be made.