Investigation on interlaminar delamination tendency of multidirectional carbon fiber composites

In this investigation carbon fiber reinforced laminates with different orientation layups are prepared and studied under tensile loading condition. Multiple strain measurement techniques, namely, resistive strain gauges, embedded optical sensors and digital image correlation are used to analyze stress-strain behavior simultaneously through the thickness of composite materials, and to determine the sequence of failure in different plies. Inconsistencies of strains measured through different methods is correlated with the tendency for interlaminar delamination, therefore demonstrating the ability of multi-instrument approach to describe damage progress through the thickness of multidirectional laminates. Complementary analysis through acoustic emission methods reveals that the angle of off-axis surface plies can influence the sequence of failure under tensile loading condition, and damage monitoring capabilities of acoustic emission system is directly affected by delamination tendency of surface plies. Remarkably, the delayed failure of off-axis plies is shown to be related to reorientation of these layer towards loading direction using infrared thermography method.     

Determination of thermal expansion coefficient of a monofilament polyamide fiber using digital image correlation

Coiled polymer actuators are a type of artificial muscles that are a promising development in the field of smart materials. The coefficient of thermal expansion of monofilament polyamide fibers is a crucial parameter for understanding the actuation of coiled fibers. The main purpose of this work is to develop a new methodology for estimating the coefficient of thermal expansion and the transition temperature of monofilament polymer fibers. In the experimental procedure, axial deformations of monofilament polyamide fiber samples were induced by temperature variations using a controlled thermal system. These deformations were determined from images of polyamide samples using the digital image correlation method. Two different approaches based on distinct temperature conditions were conducted. An alternative model with three parameters, including the coefficient of thermal expansion, was introduced to describe the thermal-mechanical behavior of monofilament polyamide fibers. Moreover, polyamide samples were also characterized using four conventional methodologies. Results indicated that the coefficient of thermal expansion changed of a modest negative value to a large negative value and this transition occurred around the glass transition temperature of the polyamide. The thermal expansion curves demonstrate good repeatability and all estimated parameters were in accordance with literature, indicating that the proposed approach can be suitable for the proposed study. This investigation may help in understanding of the intrinsic thermal-mechanical behavior of polymeric monofilaments employed as actuators.     

Dynamic fracture testing of polymethyl-methacrylate (PMMA) single-edge notched beam

Dynamic fracture in single-edge notched polymethyl-methacrylate (PMMA) beams have been investigated by three-point-bending impact testing with a drop-weight machine. A high-speed camera combined with the digital image correlation (DIC) method is used to capture the impact-induced crack initiation and propagation, as well as the beam deformation fields and the open mode strain at the original notch tip. The crack propagation length is recorded and the instantaneous crack velocity is calculated. Furthermore, the dynamic fracture toughness K_Id is quantified from the loading-displacement relations at different impact velocities. The effects of the impact velocity and impact energy on dynamic fracture toughness, fracture initiation strain, as well as the corresponding influences on the fracture propagation velocity, are discussed.     

Generalization of Strang's Preconditioner with Applications to Toeplitz Least Squares Problems

In this paper, we propose a method to generalize Strang's circulant preconditioner for arbitrary n-by-n matrices A_n. The th column of our circulant preconditioner Sn is equal to the th column of the given matrix A_n. Thus if A_n is a square Toeplitz matrix, then Sn is just the Strang circulant preconditioner. When Sn is not Hermitian, our circulant preconditioner can be defined as . This construction is similar to the forward-backward projection method used in constructing preconditioners for tomographic inversion problems in medical imaging. We show that if the matrix A_n has decaying coefficients away from the main diagonal, then is a good preconditioner for An. Comparisons of our preconditioner with other circulant-based preconditioners are carried out for some 1-D Toeplitz least squares problems: min ∥ b - Ax∥₂. Preliminary numerical results show that our preconditioner performs quite well, in comparison to other circulant preconditioners. Promising test results are also reported for a 2-D deconvolution problem arising in ground-based atmospheric imaging.     

医学影像处理的深度学习可解释性研究进展

随着医学影像数据的迅速增长,传统的影像分析方法给医生带来巨大挑战。利用计算机视觉技术提供自动或半自动辅助诊断,可大大缓解人工阅片压力,提高诊断的准确性,促进医疗流程的标准化建设等。目前,深度学习卷积神经网络在医学影像处理中已取得不俗表现,但深度学习“黑匣子”的不可解释性阻碍了智能医疗诊断的发展。为增强对医学影像数据处理的深度学习可解释性的了解,对近几年相关研究进展进行了综述。首先,综述了深度学习在医学领域的应用现状及面临的问题,对神经网络的可解释性内涵进行了讨论;然后,从现有深度学习可解释性的常见方法出发,重点讨论了医学影像处理的深度学习可解释性研究进展;最后,探讨了医学影像处理的深度学习可解释性的发展趋势。     

基于快速神经网络架构搜索的鲁棒图像水印网络算法

为解决深度学习在图像水印算法中计算量大且模型冗余的问题,提高图像水印算法在抵抗噪声、旋转和剪裁等攻击时的鲁棒性,提出基于快速神经网络架构搜索（neural architecture search,NAS）的鲁棒图像水印网络算法。通过多项式分布学习快速神经网络架构搜索算法,在预设的搜索空间中搜索最优网络结构,进行图像水印的高效嵌入与鲁棒提取。首先,将子网络中线性连接的全卷积层设置为独立的神经单元结构,并参数化表示结构单元内节点的连接,预先设定结构单元内每个神经元操作的搜索空间;其次,在完成一个批次的数据集训练后,依据神经元操作中的被采样次数和平均损失函数值动态更新概率;最后,重新训练搜索完成的网络。水印网络模型的参数量较原始网络模型缩减了92%以上,大大缩短了模型训练时间。由于搜索得到的网络结构更为紧凑,本文算法具有较高的时间性能和较好的实验效果,在隐藏图像时,对空域信息的依赖比原始网络更少。对改进前后的2个网络进行了大量鲁棒性实验,对比发现,本文算法在CIFAR-10数据集上对抵抗椒盐噪声和旋转、移除像素行（列）等攻击优势显著;在ImageNet数据集上对抵抗椒盐高斯噪声、旋转、中值滤波、高斯滤波、JPEG压缩、裁剪等攻击优势显著,特别是对随机移除行（列）和椒盐噪声有较强的鲁棒性。     

基于神经网络的三维宽场显微图像复原研究

提出一种利用BP神经网络进行三维宽场显微图像复原的非线性映射方法,将三维图像转化为二维图像进行处理,利用神经网络的学习能力,通过训练,建立含有散焦信息的二维模糊图像与二维清晰图像之间的映射关系,然后对切片堆叠进行逐幅复原,从而实现显微图像的三维复原.得到的复原图像在视觉上和定量分析上都获得了很好的效果.由于采用小规模神经网络,训练时间短,计算量小,使实时复原成为可能.     

基于灰度冗余的红外图像自适应输出窗技术

根据红外图像的特点,提出了一种基于灰度冗余的红外图像自适应输出窗技术.该技术通过对图像灰度直方图的统计,设定适当的阈值,将灰度像素分布为零的灰度级进行完全压缩,将小于阈值的灰度级映射到阈值灰度级,然后将有效灰度级在整个可显示灰度级范围内作等间距排列.该算法在压缩灰度冗余的同时,实现了对有效图像灰度级的无损均衡,提高了图像质量,并能实现实时处理.     

基于改进U-Net++的CT影像肺结节分割算法

卷积神经网络的语义分割模型未有效利用特征权重信息,导致在医学图像复杂场景中分割边界出现欠分割现象.针对该问题,基于融合自适应加权聚合策略提出一种改进的U-Net++网络,并将其应用于电子计算机断层扫描影像肺结节分割.该模型首先在卷积神经网络中提取出不同深度特征语义级别的信息,再结合权重聚合模块,自适应地学习各层特征的权...     

一种基于HSI颜色模型的目标提取方法

在工业、农业、军事以及医疗等领域中,目标识别是一个关键技术,近年来在图像处理方面已成为一个重要的研究课题。提出了一种用于除草机器人的基于HSI(色调、明度和饱和度)颜色模型的彩色图像目标提取方法。该方法首先将具有某种色彩信息的杂草与农作物从背景土壤中分离出来,然后再根据杂草与农作物形状的不同,采用腐蚀和膨胀的方法进一步将杂草与农作物分离。该方法除了用于除草机器人以外,还可用于水果采摘等。仿真结果验证了该方法的有效性。