Cs-K混合蒸气中Cs(8D)+K(4S)碰撞能量转移

在Cs-K混合蒸气中，两步激发Cs原子到8D态，观察了Cs(8D) K(4S)→Cs(5D) K(4P)碰撞能量合并逆过程(REP，reverse energy pooling)。应用双调制技术探测K(4P)原子发射的荧光，基态K原子密度用光学吸收方法测量。得到了REP速率系数，讨论了其它过程对速率系数的影响．     

深度连续卷积神经网络模型构建与性能分析

为了提升卷积神经网络特征提取能力,设计了一种基于连续卷积的深度卷积神经网络模型.该模型采用小尺度的卷积核来更细致地提取局部特征,并借助连续的两个卷积层增加模型的非线性表达能力,结合Dropout技术降低神经元之间的相互依赖,利用抑制网络过拟合对模型进行优化.人脸表情、手写数字字符和彩色图像的目标识别实验表明,在图像较为复杂时,该模型在识别的准确性和泛化性能上比手工特征提取方法及一般的2、3层卷积结构具有明显的优势.     

多线程管理下的数据库访问技术

对数据库服务器访问的管理,是提高C／S架构系统性能的重要环节。采用多线程技术对即时访问通道进行动态管理,可有效降低系统资源消耗,提高系统响应速度。     

基于视角置信度和注意力的暴力行为识别

暴力行为容易出现遮挡情况, 识别准确率较低. 目前, 一些算法加入多视角视频输入来解决遮挡问题, 以等量权重将所有视角数据融合, 但是不同视角的视频因拍摄距离和遮挡情况本身就对识别存在差异性. 针对该问题, 本文提出一种基于视角置信度和注意力的暴力行为识别方法, 提高暴力识别的准确率. 本文将时序差分模块TDM的输入扩展成多视角, 将通道注意力机制运用在片段维度来增强TDM中跨段特征提取能力, 通过背景抑制方法突显移动目标的纹理特征并计算出每个视角图像的置信度, 引入双线性池化方法融合多视角视频特征, 根据视角置信度分配每个视角局部特征的权重. 本文在公开数据集CASIA-Action和自制数据集上进行了验证. 实验表明, 本文提出的视角置信度方法优于改进前的双线性池化方法, 暴力行为准确率相较于现有的行为识别方法取得了更好的效果.     

基于超像素级卷积神经网络的多聚焦图像融合算法

该文提出了基于超像素级卷积神经网络(sp-CNN)的多聚焦图像融合算法。该方法首先对源图像进行多尺度超像素分割,将获取的超像素输入sp-CNN,并对输出的初始分类映射图进行连通域操作得到初始决策图;然后根据多幅初始决策图的异同获得不确定区域,并利用空间频率对其再分类,得到阶段决策图;最后利用形态学对阶段决策图进行后处理,并根据所得的最终决策图融合图像。该文算法直接利用超像素分割块进行图像融合,其相较以往利用重叠块的融合算法可达到降低时间复杂度的目的,同时可获得较好的融合效果。     

CCNet: CNN model with channel attention and convolutional pooling mechanism for spatial image steganalysis

Image steganalysis based on convolutional neural networks(CNN) has attracted great attention. However, existing networks lack attention to regional features with complex texture, which makes the ability of discrimination learning miss in network. In this paper, we described a new CNN designed to focus on useful features and improve detection accuracy for spatial-domain steganalysis. The proposed model consists of three modules: noise extraction module, noise analysis module and classification module. A channel attention mechanism is used in the noise extraction module and analysis module, which is realized by embedding the SE(Squeeze-and-Excitation) module into the residual block. Then, we use convolutional pooling instead of average pooling to aggregate features. The experimental results show that detection accuracy of the proposed model is significantly better than those of the existing models such as SRNet, Zhu-Net and GBRAS-Net. Compared with these models, our model has better generalization ability, which is critical for practical application.     

Cost benefits from standardization of the packaging glass bottles

This study deals with a recycling system with two competing brewers. It is assumed that they coordinate their manufacturing operations through standardization of their glass bottles for easy implementation of extended producer responsibility (EPR). Immediate benefits from the standardization are three folds. Firstly, the sorting and exchange processes of the bottles collected for reuse by each brewer become no longer necessary. Secondly, cost reduction is achieved through streamlining of collection and reuse processes. Finally, under the stochastic demand of glass bottles their inventory holding costs and lost sales cost are reduced via inventory pooling. Through the development of the mathematical models we determine an optimal operation policy of the two brewers that maximizes the sum of benefits obtained from standardization. Numerical examples are solved to show the validity of the model. Sensitivity tests are also performed to examine the effects of system parameters on the objective function value and decision variables.     

基于J2EE的专利检索系统设计与优化

文中介绍了J2EE的技术框架，讨论了J2EE的分层结构和企业服务；基于该技术框架，提出了一个适合中国专利文献检索和管理的体系结构，设计并实现了该系统的基本功能模块；最后提出来了一种基于组件池的系统优化算法，并给出了相应的试验结果。     

基于分层压缩激励的ASPP网络单目深度估计

场景深度估计是场景理解的一项基本任务,其准确率反映了计算机对场景的理解程度.传统的深度估计利用金字塔池化(ASPP)模块可以在不改变图像分辨率的情况下处理不同像素特征,但该模块未考虑不同像素特征之间的关系,导致场景特征提取不准确.针对ASPP模块在深度估计中出现的弊端,提出了一种改进型的ASPP模块,解决了该模块在图像...     

基于稀疏自编码深度神经网络的林火图像分类

针对林火与相似目标很难区分的问题,提出一种基于稀疏自编码深度神经网络的林火图像分类新方法。采用无监督的特征学习算法稀疏自编码从无标签图像小块中学习特征参数,完成深度神经网络的训练;利用学习到的特征从原始大小分类图像中提取特征并卷积和均值池化特征;对卷积和池化后的特征采用softmax回归来训练最终softmax分类器。实验结果表明,跟传统的BP神经网络相比,新方法能够更有效区分林火与红旗、红叶等类似物体。