基于统计特征和马尔可夫特征的图像拼接盲检测

数字图像真实性检测在司法鉴定等领域有着重要的作用.常见的图像拼接篡改会降低图像像素直接的相关性,这可以通过一些统计特征反映出来.采用特征提取-分类的方法,提取矩特征、基于二维相位一致性的统计特征,结合DCT域的马尔可夫特征,利用SVM分类器进行分类,实现了拼接图像的盲检测.实验结果表明,该方法有较好的鉴别准确率,可达91.75%.     

有机磷水解酶生物传感器载体固定化酶的研究

有机磷水解酶（OPH）传感器作为检测农产品中农药残留的新型检测装置,其酶的固定化对OPH传感器的灵敏度和稳定性有重要的影响。研究了几种酶固定化载体、孔径大小、固定方式、固定方法（试剂组成）对传感器pH值的影响。结果显示：采用孔径为0.45μm的硝酸纤维素膜制备固定化酶片的pH值要大于其余几种;采用浸泡方式制备固定化酶片的pH值明显大于传统的滴定法;采用牛血清白蛋白（BSA）、戊二醛交联固定的效果优于酶直接吸附法和BSA固定法,且当戊二醛体积分数为2.5％,BSA为10％时,酶固定化效果最好。     

基于蕴涵算子族L-λ-R_0的模糊推理α-三I约束算法

提出了基于蕴涵算子族L-λ-R0的模糊推理的思想,这将有助于提高推理结果的可靠性。针对蕴涵算子族L-λ-R0给出了模糊推理的FMP模型及FMT模型的α-三I约束算法。     

基于蕴涵算子族Lλ0λG的反向三I支持算法

提出了基于蕴涵算子族Lλ0λG的模糊推理的思想,这将有助于提高推理结果的可靠性。针对蕴涵算子族Lλ0λG给出了模糊推理的FMP模型及FMT模型的反向三I支持算法。     

融合MAML和CBAM的安卓恶意应用家族分类模型

为满足对新兴安卓恶意应用家族的快速检测需求,提出一种融合MAML(model-agnostic meta-learning)和CBAM(convolutional block attention module)的安卓恶意应用家族分类模型MAML-CAS。将安卓恶意应用样本集中的DEX文件可视化为灰度图,并构建任务集;融合混合域注意力机制CBAM,设计两个具有同等结构的卷积神经网络,分别作为基学习器和元学习器,这两个学习器在自动提取任务集中样本特征的同时,可从通道和空间两个维度来增强关键特征表达;利用元学习方法 MAML对两个学习器进行训练,其中基学习器完成特定恶意家族分类任务的属性学习,元学习器则学习不同任务的共性;在两个学习器训练完成后,MAML-CAS将获得初始化参数,在面对新的安卓恶意应用家族分类任务时,不需要重新训练,只需要少量样本就可以快速迭代;利用训练完成的基学习器提取安卓恶意应用家族特征,并利用SVM进行恶意家族分类。实验结果表明,MAML-CAS模型对新兴小样本安卓恶意应用家族具有良好的检测效果,检测速度较快,并具有较好的稳定性。     

Joint decision of product configuration and remanufacturing for product family design

Product family design (PFD) is a popular method for increasing product variety to satisfy the needs of diversified markets. With the increasing concern for environmental friendliness in society, more and more companies develop launching remanufactured products and include them in their product families. Therefore, PFD should be considered in a broad decision space where configuration of product variants and remanufacturing are considered simultaneously. However, this issue was not addressed properly in previous research. In this paper, a methodology for joint decision of product configuration and remanufacturing is proposed in which a bi-objective mixed integer programming model is formulated to determine the configurations of both new and remanufactured products for minimising product cost, maximising total market share, and satisfying reliability requirements. Then, Non-dominated Sorting Genetic Algorithm II (NSGAII) is adopted to solve the optimization problem. Computational experiments were conducted and their results show that NSGAII is convergent to the model well. A case study is presented to illustrate the applicability and effectiveness of the proposed methodology.     

AZ31镁合金室温拉伸微观变形机制EBSD原位跟踪研究

利用电子背散射衍射(EBSD)技术,原位跟踪AZ31镁合金轧制板材室温下沿轧向拉伸时的晶粒取向变化。对变形过程的滑移系和孪晶启动机进行分析。结果表明:变形过程主要由〈a〉基面和柱面滑移系开动而实现,晶粒取向无明显变化,大量〈a〉位错滑移的产生,使得变形后小角度晶界增加明显。晶粒中拉伸孪晶是试样在拉伸变形过程中产生的,而非在试样拉伸后的卸载过程中产生。     

通过预涂覆SiC优化AZ31B镁合金DE-GMAW焊缝的显微组织及提升力学性能

研究了碳化硅(SiC)颗粒涂覆量对2mm厚AZ31B镁合金双电极气体保护焊DE-GMAW焊缝的宏观形貌、显微组织和力学性能的影响。利用光学显微镜、扫描电镜和X射线衍射分析技术(XRD)对接头的显微组织、相及相的成分进行分析。同时采用维氏硬度试验计和万能试验机对焊接接头的显微硬度和抗拉强度分别进行测量。结果表明,随着SiC涂覆量的增加,焊接接头熔深和深宽比先增大后减小。XRD测试结果显示,SiC的添加并不改变焊缝中相的组成,焊缝由α-Mg和β-Mg_(17)Al_(12)组成。SiC可以细化α-Mg晶粒,打断焊缝中β-Mg_(17)Al_(12)相,起到弥散强化的效果。但SiC涂覆量达到一定值后,随着SiC涂覆量的继续增加,α-Mg晶粒粗化,β-Mg_(17)Al_(12)相的弥散强化作用不再增加。焊接接头的显微硬度和抗拉强度随着SiC涂覆量的增加先增大后减小。     

Miniversal deformations of linear systems under the full group action

We consider the equivalence relation in the space of time-invariant linear dynamical systems of the form

under the full group action of state feedback/output injection transformations. As in the case of the reduction of a matrix to its Jordan canonical form, the reduction of a quadruple (A,B,C,D) defining a system as above to its canonical reduced form is an unstable operation. Following V.I. Arnold’s techniques, the starting point for the study of local perturbations is to obtain a miniversal deformation of a differentiable family of quadruples. In this paper, a “real” miniversal deformation and some applications are shown.     

Functional and metabolic evaluation of the hypertrophied heart using MRI and31P-MRS

4. Conclusions Diastolic LV function and myocardial HEP metabolism are impaired only when LVH is caused by permanent pressure or volume overload, and not by a temporary increase in cardiac workload during part of the day as in elite athletes. Therefore, training-induced and pressure/volume-overload-induced LVH seem to represent different phenotypes of LVH, possibly related to genetic reprogramming which only occurs during permanent cardiac overload [17]. Moreover, there is an association between impaired LV diastolic function and altered myocardial HEP metabolism in patients with hypertension and in patients with aortic valve disease. Finally we did not find a correlation between myocardial HEP metabolism and LV mass in any of the groups studied. The latter indicates that LVH should be regarded as an epiphenomenon to cardiac overload, and not as a primary factor causing abnormal HEP metabolism.