基于知识的海杂波背景下距离扩展目标检测

该文针对在辅助数据有限的情况下自适应检测器出现检测性能损失,提出基于杂波的先验知识分布的距离扩展目标自适应检测算法。复合高斯杂波的纹理和散斑的协方差矩阵分别被建模为服从逆伽玛分布的随机变量和逆复Wishart分布的随机矩阵。利用先验知识推导了纹理分量的最大后验估计,并结合广义似然比检验设计了不依赖辅助数据的距离扩展目标自适应检测器。仿真结果表明,提出的检测器在参数失配条件下具有好的鲁棒性,而且在辅助数据有限情况下检测性能优于传统的广义似然比检测器的检测性能。     

A novel etchant for revealing the prior austenite grain boundaries and matrix information in high alloy steels

To observe the prior austenite grain (PAG) boundaries in high-alloy steels, a novel etchant composed of oxalic acid, hydrogen peroxide, sodium bisulfite, and water has been developed. Etching with this system successfully revealed both the grain boundaries and the deformation bands within grains in high Co–Ni martensitic alloy steels. The carbon content of the alloys used in the test program varied from 0.23% to 0.35%, the cobalt content was in the range 9–13%, and the nickel content was in the range 8–11%. The specimens were austenitized at temperatures in the range 900–1200 °C. The rolling temperatures ranged from 700 °C to 1000 °C. The grain sizes in these alloys and those containing titanium were clearly revealed by optical microscopy following application of the new etchant.     

投资监理在前期费用控制中的作用

工程的前期费用具有难以预测、费用种类多及计费规则复杂等特点。本文从工程实际出发,对投资监理在前期费用控制中应起到的作用进行探讨和研究,认为投资监理作为对工程投资控制实施专业化管理的参建者,必须以工程投资者的利益为导向,以自身丰富的工作经验为业主服务,确保工程前期费用的投资控制安全,进而实现整个工程的投资控制目标。     

Statistical analysis of the effects of prior load on fracture

Collated fracture data for three steels, A533B, A508 and BS1501, and an Aluminium alloy, 2650, were examined to assess the statistical significance of the effect of prior loading on subsequent fracture. Weibull statistical and probabilistic analysis was used throughout. Two prior loading conditions were examined; one associated with warm pre-stress, and a second using pre-compression. For the former, prior loading resulted in an increase in the mean toughness together with an increase in the shape parameter and a decrease in variability compared to the as-received material. In contrast, when prior loading involved out-of-plane compressive loading (or side-punching) statistical evidence revealed that there was a reduction in toughness together with a decrease in the shape parameter and an increase in variability. Two numerical models were applied to steel data and were able to predict the overall trends obtained from the experiments, but could not reproduce accurately the experimental statistical distributions.     

基于生成对抗网络的图像去雾算法

与基于图像增强的去雾算法和基于物理模型的去雾算法相比,基于深度学习的图像去雾方法在一定程度上提高计算效率,但在场景复杂时仍存在去雾不彻底及颜色扭曲的问题.针对人眼对全局特征和局部特征的感受不同这一特性,文中构建基于生成对抗网络的图像去雾算法.首先设计多尺度结构的生成器网络,分别以全尺寸图像和分割后的图像块作为输入,提取图像的全局轮廓信息和局部细节信息.然后设计一个特征融合模块,融合全局信息和局部信息,通过判别网络判断生成无雾图像的真假.为了使生成的去雾图像更接近对应的真实无雾图像,设计多元联合损失函数,结合暗通道先验损失函数、对抗损失函数、结构相似性损失函数及平滑L1损失函数训练网络.在合成数据集和真实图像上与多种算法进行实验对比,结果表明,文中算法的去雾效果较优.     

引入低秩约束先验的深度子空间聚类

大多数子空间聚类算法将高维数据映射到低维子空间时不能较好捕获数据间几何结构.针对上述问题,文中提出引入低秩约束先验的深度子空间聚类算法,兼顾数据全局和局部结构信息.算法结合低秩表示与深度自编码器,利用低秩约束捕获数据全局结构,并将约束神经网络的潜在特征表示为低秩.自编码通过最小化重构误差进行非线性低维子空间映射,保留数据的局部特性.以多元逻辑回归函数作为判别模型,预测子空间分割.整个算法在无监督联合学习框架下进行优化.在5个数据集上的实验验证文中方法的有效性.     

基于狄利克雷过程混合模型的内外先验融合

近年来,使用高斯混合模型作为块先验的贝叶斯方法取得了优秀的图像复原性能,针对这类模型分量固定及主要依赖外部学习的缺点,提出了一种新的基于狄利克雷过程混合模型的图像先验模型。该模型从干净图像数据库中学习外部通用先验,从退化图像中学习内部先验,借助模型中统计量的可累加性自然实现内外部先验融合。通过聚类的新增及归并机制,模型的复杂度随着数据的增大或缩小而自适应地变化,可以学习到可解释及紧凑的模型。为了求解所有隐变量的变分后验分布,提出了一种结合新增及归并机制的批次更新可扩展变分算法,解决了传统坐标上升算法在大数据集下效率较低、容易陷入局部最优解的问题。在图像去噪及填充实验中,相比传统方法,所提模型无论在客观质量评价还是视觉观感上都更有优势,验证了该模型的有效性。     

煮糖过程的新型建模方法

煮糖过程的主要非线性表现为晶体生长率。本论文采用两种方法来描述晶体生长率模型的设计，第1种方法是通过采用非线性规划最优化方法(NLP)决定晶体生长率经验公式的参数；第2种方法是一种新颖的建模策略，它把晶体生长率作为非线性逼近器的人工神经网络(ANN)与由蔗糖晶体质量平衡所表述的先前既定的知识相结合。最初结果显示第1种类型的模型能够较好地执行局部匹配，而第2种具有更大的灵活性。     

Spatially constrained ICA algorithm with an application in EEG processing

Independent Component Analysis (ICA) aims at blindly decomposing a linear mixture of independent sources. It has lots of applications in diverse research areas. In some applications, there is prior knowledge on the sources and/or the mixing vectors. This prior knowledge can be incorporated in the computation of the independent sources. In this paper we provide an algorithm for so-called spatially constrained ICA (scICA). The algorithm deals with the situation when one mixing vector is exactly known. Also the generalization to more mixing vectors is discussed. Numerical experiments are reported that allow us to assess the improvement in accuracy that can be achieved with these algorithms compared to fully blind ICA and to a previously proposed constrained algorithm. We illustrate the approach with a biomedical application.     

融合先验形状知识的主动轮廓模型研究

融合先验形状知识的主动轮廓模型（HACM）能够解决目标存在遮挡、局部形变、相似变换等实际问题,广泛应用于图像分割、轮廓提取等领域。本文以主动轮廓模型（ACM）为基础,从先验形状知识的提取以及知识的融合方法两个方面,总结近年来ACM在融合先验形状知识的目标轮廓提取领域的一些最新研究成果。首先研究基于区域不变矩描述子、样条函数以及水平集的形状知识提取方法,用于提取刚体性、非刚体性目标样本集所包含的轮廓信息,该信息具有一定的泛化能力,即针对样本集以外的目标轮廓具备一定的表达能力。然后以单目标轮廓提取问题为研究对象,从距离函数的角度分析各种先验形状知识与原始轮廓模型的全局性融合方法,并介绍了一种基于标识函数的局部融合方法。局部融合方法可以解决单知识、多目标情况下的轮廓提取问题。最后对文中所介绍模型进行总结分析,对该领域的研究方向进行了展望。