基于循环互相关系数的CGAN温度值图像扩增

针对变电设备红外图像温度值样本少、不均衡等问题,本文提出了一种基于循环互相关系数的条件生成对抗网络(conditional generative adversarial network, CGAN)温度值图像扩增方法。根据图像相似度提出了循环互相关系数,改进了CGAN模型的损失函数;使用改进后的CGAN模型在原始温度值图像数据集的基础上进行图像扩增,得到了包含11种标签的新数据集;采用卷积神经网络对传统图像扩增方法、原始CGAN模型和改进的CGAN模型扩增的图像进行对比和测试。结果表明,改进的CGAN模型收敛速度更快,训练过程稳定,扩增的图像轮廓清晰、细节丰富,客观评价指标最高,温度值识别准确率达到99.4%,实现了图像扩增的目的。     

Staphylococcus aureus Infection-Related Glomerulonephritis with Dominant IgA Deposition

Since 1995, when we reported the case of a patient with glomerulonephritis with IgA deposition that occurred after a methicillin-resistant Staphylococcus aureus (MRSA) infection, many reports of MRSA infection-associated glomerulonephritis have accumulated. This disease is being systematized as Staphylococcus infection-associated glomerulonephritis (SAGN) in light of the apparent cause of infection, and as immunoglobulin A-dominant deposition infection-related glomerulonephritis (IgA-IRGN) in light of its histopathology. This glomerulonephritis usually presents as rapidly progressive glomerulonephritis or acute kidney injury with various degrees of proteinuria and microscopic hematuria along with an ongoing infection. Its renal pathology has shown several types of mesangial and/or endocapillary proliferative glomerulonephritis with various degrees of crescent formation and tubulointerstitial nephritis. IgA, IgG, and C₃ staining in the mesangium and along the glomerular capillary walls have been observed on immunofluorescence examinations. A marked activation of T cells, an increase in specific variable regions of the T-cell receptor β-chain-positive cells, hypercytokinemia, and increased polyclonal immune complexes have also been observed in this glomerulonephritis. In the development of this disease, staphylococcal enterotoxin may be involved as a superantigen, but further investigations are needed to clarify the mechanisms underlying this disease. Here, we review 336 cases of IgA-IRGN and 218 cases of SAGN.     

融合表情和BVP生理信号的双模态视频情感识别

目的 针对当前视频情感判别方法大多仅依赖面部表情、而忽略了面部视频中潜藏的生理信号所包含的情感信息,本文提出一种基于面部表情和血容量脉冲（BVP）生理信号的双模态视频情感识别方法。方法 首先对视频进行预处理获取面部视频;然后对面部视频分别提取LBP-TOP和HOG-TOP两种时空表情特征,并利用视频颜色放大技术获取BVP生理信号,进而提取生理信号情感特征;接着将两种特征分别送入BP分类器训练分类模型;最后利用模糊积分进行决策层融合,得出情感识别结果。结果 在实验室自建面部视频情感库上进行实验,表情单模态和生理信号单模态的平均识别率分别为80%和63.75%,而融合后的情感识别结果为83.33%,高于融合前单一模态的情感识别精度,说明了本文融合双模态进行情感识别的有效性。结论 本文提出的双模态时空特征融合的情感识别方法更能充分地利用视频中的情感信息,有效增强了视频情感的分类性能,与类似的视频情感识别算法对比实验验证了本文方法的优越性。另外,基于模糊积分的决策层融合算法有效地降低了不可靠决策信息对融合的干扰,最终获得更优的识别精度。     

Effects of Solvents in the Polyethylene Glycol Series on the Bonding of Copper Joints Using Ag2O Paste

The effects of reducing solvents on the bonding process using silver oxide paste in a copper joint were investigated. Three solvent types were tested: diethylene glycol (DEG), triethylene glycol (TEG), and polyethylene glycol (PEG). The strength of the joints was assessed by fracturing, which occurred at the interface of the copper oxide layer and the copper substrate in DEG and TEG samples and at the bonded interface in the PEG sample. Analysis of the samples revealed that, in the DEG and TEG samples, the copper substrate was oxidized during the bonding process, which compromised the shear strength of the joints. In contrast, the PEG sample exhibited nonuniform sintering of the silver layer while retaining good shear strength. It was found that the combination of DEG and PEG produced optimum shear strength in the copper joint, as PEG suppressed the growth of copper oxide and DEG promoted the formation of a dense sintered silver layer. The bonding strength achieved was higher than that of the gold-to-gold joint made using standard Pb-5Sn solder.     

基于三态信号的改进游程编码压缩方法

为提高集成电路测试效率,提出一种结合三态信号的改进游程编码压缩方法.先对原始测试集进行部分输入精简处理并填充测试集的无关位,再对经过预处理的测试集根据游程长度进行变长分段处理找出最优段长.按照游程长度的出现频率对最优段长下的参考位设置编码表进行编码压缩,使用三态信号编码标志位并将编码压缩后的测试集存入自动测试设备(AT...     

Image blurring of thick specimens due to MeV transmission electron scattering: a Monte Carlo study

Image blurring of MeV transmission electrons for gold nanoparticles on the top surface of micrometer-thick specimens has been investigated using the Monte Carlo simulation. Both the simulated line density profile and therefore image blurring were in good agreement with the experimental ones in the ultrahigh voltage electron microscope. Quantitative effects of specimen thickness and electron energy on image blurring were presented, in which the specimen thickness had a greater influence. Image blurring was demonstrated to be caused mainly by multiple elastic scattering, but it could be reduced to several nanometers for a 5?μm thick epoxy-resin specimen at the electron energy of 2?MeV.     

Adhesion between polymer surface modified by graft polymerization and tissue during surgery using an ultrasonically activated scalpel device

In the field of surgery, achieving adhesion between a polymer implant and tissue poses a challenge considering that suturing is not appropriate for the stability of such implants. An ultrasonically activated scalpel that generates heat by mechanical vibration and promotes adhesion between a polymer implant and native tissue by pressing the two materials together has very good potential for application in the field. To determine the type of polymer that is suitable for the purpose, we investigated polyethylene (PE) and polystyrene (PS) films, the surfaces of which were activated by corona discharge. Graft polymerization was then performed on the corona‐treated surfaces to vary their properties. The corona‐treated PE and PS films grafted with poly(acrylic acid), poly(methacrylic acid), poly(vinyl benzylacrylic acid), and poly(hydroxylethyl acrylate), respectively, adhered to the tissue when the ultrasonically activated scalpel was applied. The heat generated by the mechanical vibration and the applied pressure enabled the carboxyl or hydroxyl groups to bond with the proteins in the extracellular matrix. We therefore concluded that it was possible to integrate this technique in the development of new types of polymer devices that could be stably implanted in a living body. © 2014 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2014 , 131, 40885.     

Self-swarming robots that exploit hydrodynamical interaction

Although numerous swarm robotic systems have already been developed, they have exhibited limited adaptability. This was partly because the previous researchers designed the control schemes on the basis of informational interaction, without considering the physical effects of the environment. To tackle this problem, we employ an unconventional approach: we design a control scheme for swarm robots based on their physical interaction in a hydrodynamic field, inspired by biological swarming. Through simulations using a smoothed particle hydrodynamics method, we show that the proposed control scheme allows agents to form an ordered swarm in response to their environment.     

Piezoresistivities of vapor‐grown carbon fiber/silicone foams for tactile sensor applications

Due to the growing demand for tactile sensors, the possibility of detecting an external uniaxial pressure by the piezoresistive measuring of a conductive filler/elastomer composite was investigated. A series of piezoresistive models are discussed. Novel highly sensitive piezoresistive foams with excellent elasticity were fabricated using vapor‐grown carbon fiber (VGCF), two‐component silicone elastomer and a new type of thermally expandable micro beads foaming agent to overcome the disadvantages of the silicone elastomer in the utilization of a tactile sensor. Deformations of the foams caused by uniaxial pressure were observed using scanning electron microscopy from cross‐sections. Effects of the VGCF and the foaming agent on the piezoresistivitiy were investigated. The piezoresistive mechanisms of the foams are discussed according to the measurements, and good fit was found between the theoretical calculations and the experimental piezoresistivity measurements. It is found that the addition of the micro beads foaming agent can improve the piezoresistivity of the VGCF/silicone foam and increase the sensitivity and repeatability for its application in a tactile sensor. © 2016 Society of Chemical Industry     

Multi-classifier ensemble based on dynamic weights

Multimedia Tools and Applications - In this study, a novel multi-classifier ensemble method based on dynamic weights is proposed to reduce the interference of unreliable decision information and...