基于全局与序列变分自编码的图像描述生成

基于Transformer架构的图像描述生成方法通常学习从图像空间到文本空间的确定性映射,以提高预测“平均”描述语句的性能,从而导致模型倾向于生成常见的单词和重复的短语,即所谓的模式坍塌问题。为此,将条件变分自编码与基于Transformer的图像描述生成相结合,利用条件似然的变分证据下界分别构建了句子级和单词级的多样化图像描述生成模型,通过引入全局与序列隐嵌入学习增强模型的隐表示能力。在MSCOCO基准数据集上的定量和定性实验结果表明,两种模型均具备图像到文本空间的一对多映射能力,相比于目前最新的方法COS-CVAE（diverse image captioning with context-object split latent spaces）,在随机生成20个描述语句时,准确性指标CIDEr和多样性指标Div-2分别提升了1.3和33%,在随机生成100个描述语句的情况下,CIDEr和Div-2分别提升了11.4和14%,所提方法能够更好地拟合真实描述分布,在多样性和准确性之间取得了更好的平衡。     

密文图像中的可逆信息隐藏算法

针对传统算法存在的加密和信息隐藏过程分离,嵌入容量小,操作不灵活的缺陷,提出了一种基于混沌和直方图平移的密文图像中的可逆信息隐藏算法。发送端先对图像加密,然后嵌入信息;接收端既可以先解密再提取信息,也可以先提取信息再解密。仿真实验结果表明该算法易于实现,并能有效提高信息嵌入的容量。     

High-dimensional image data feature extraction by double discriminant embedding

We propose a supervised feature extraction method in this paper that uses two successive transformations to produce the extracted features. The first projection maximizes the difference between spectral features. Thus, produced features have minimum overlap in the new feature space. The second projection maximizes the discrimination between classes. The proposed method, which is called double discriminant embedding (DDE), uses just the first statistics of data. Thus, DDE has good efficiency using limited training samples. The experimental results on four popular hyperspectral images show the better efficiency of DDE in comparison with LDA, GDA, NWFE, and supervised LPP methods in small sample size situation.     

Algebraic notions of nontermination: Omega and divergence in idempotent semirings

Two notions of nontermination are studied and compared in the setting of idempotent semirings: Cohen’s omega operator and a divergence operator. They are determined for various computational models, and conditions for their existence and their coincidence are given. It turns out that divergence yields a simple and natural way of modelling infinite behaviours of programs and discrete systems, whereas the omega operator shows some anomalies.     

基于超球体多类支持向量数据描述的医学图像分类新方法

针对乳腺X光医学图像多分类问题中训练速度比较慢的问题,提出超球体多分类支持向量数据描述(HSMC-SVDD)分类算法,即把超球体单分类支持向量数据描述直接扩展到超球体多分类支持向量数据描述。通过对乳腺X光图像提取灰度共生矩阵特征;然后用核主成分分析(KPCA)对数据进行降维;最后用超球体多分类支持向量数据描述分类器进行分类。由于每一类样本只参与构造一个超球体的训练,因此训练速度明显提高。实验结果表明,这种超球体多分类支持向量数据描述分类器的平均训练时间为21.369s,训练时间比Wei等(WEI L Y, YANG Y Y, NISHIKAWA R M,et al.A study on several machine-learning methods for classification of malignant and benign clustered micro-calcifications. IEEE Transactions on Medical Imaging, 2005, 24(3): 371-380)提出的组合分类器（平均训练时间40.2s）减少了10~20s,分类精度最高达76.6929%,适合解决类别数较多的分类问题。     

基于分层嵌入的二值图像隐写方法

提出了一种非加性失真的二值图像隐写方法,该方法对载体图像进行分割,生成两张子图,通过传递影响因子更新失真,并使用最小化失真隐写编码实现消息嵌入。相较于固定失真不变的加性失真隐写方法,所提方法结合了二值图像只有黑色和白色两种像素的特性,使其能够捕捉像素点间修改的相关性,实现动态修改失真,从而提高了二值图像隐写的视觉质量和隐写安全性。     

基于数据仓库技术的消防分析系统设计研究

本文以消防业务数据为基础,讨论了消防分析系统中数据仓库的设计方法、特点和优势,并对数据仓库建模中的各个关键步骤和要点进行了相关阐述.研究结果表明,利用数据仓库技术,在一定程度上对消防相关部门的分析决策起到了辅助作用.     

基于描述方法的SAR图像分割*

针对减少SAR图像分割中自由参数的问题,提出了基于最小描述长度的SAR图像分割方法。该方法经对数变换将SAR图像乘性噪声转换为加性噪声,对其建立描述模型,在描述长度最短意义上计算出重建图像,在假设SAR图像各区域实际地物后向散射特性对应的像素值恒定的前提下,该重建图像即为SAR图像的分割结果。该方法在分割的同时很好地抑制了SAR图像的相干斑噪声,保留了原始SAR图像的区域边界,并且不需要参数调节,整个分割过程自动完成,是一种非监督SAR图像分割方法。给出了该方法的具体实现步骤,实验结果验证了该方法的有效性。     

Algebraic implementation of abstract data types

Starting with a review of the theory of algebraic specifications in the sense of the ADJ-group a new theory for algebraic implementations of abstract data types is presented.While main concepts of this new theory were given already at several conferences this paper provides the full theory of algebraic implementations developed in Berlin except of complexity considerations which are given in a separate paper. The new concept of algebraic implementations includes implementations for algorithms in specific programming languages and on the other hand it meets also the requirements for stepwise refinement of structured programs and software systems as introduced by Dijkstra and Wirth. On the syntactical level an algebraic implementation corresponds to a system of recursive programs while the semantical level is defined by algebaic constructions, called synthesis, restriction and identification. Moreover the concept allows composition of implementations and a rigorous study of correctness. The main results of the paper are different kinds of correctness criteria which are applied to a number of illustrating examples including the implementation of sets by hash-tables. Algebraic implementations of larger systems like a histogram or a parts system are given in separate case studies which, however, are not included in this paper.     

Virtual transformations of data structures

Conclusion The initial push to this research is probably due to Dahl and Hoare's Hierarchical Program Structures [6] which deeply influenced the author at the time.Translated from Kibernetika, No. 1, pp. 62–67, January–February, 1980.     

Computer simulations of association-based image search mechanisms basing on theory of active perception

We discuss an approach to development of an associative memory model from the viewpoint of the theory of active perception. The theory of active perception allows one to develop the treebased memory model without the defects of the kd-tree and the vp-tree. Applications of the proposed model for solving problems of an image search by content from database are described. Also, we present the results of computer simulations directed at searching of similar and distorted images.     

基于差分扩展的高效数字图像嵌入算法研究

针对某些特殊情况需要将数据进行安全可靠传输的问题,本文提出一种基于差分扩展的将数据嵌入数字图像的算法。即将数字图像作为待传输数据的载体,按照差分扩展原理将待传数据嵌入到数字图像后进行发送,再由接收方按照与之相逆的方法进行数据恢复。经过实验表明,该算法与其它算法比较,在保证图像有很低的失真率的同时还确保更高的数据嵌入率     

A survey and taxonomy on virtual data center embedding

The Journal of Supercomputing - Data center network virtualization is being considered as a promising technology to provide a performance guarantee for cloud computing applications. One important...     

Algebraic break of image ciphers based on discretized chaotic map lattices

In this paper, we provide an algebraic cryptanalysis of a recently proposed chaotic image cipher. We show that the secret parameters of the algorithm can be revealed using chosen-plaintext attacks. Our attack uses the orbit properties of the permutation maps to deduce encryption values for a single round. Once a single round encryption is revealed, the secret parameters are obtained using simple assignments.     

基于数据点松紧度的局部线性嵌入算法

局部线性嵌入算法（LLE）是流形学习中非线性数据降维的重要方法之一。考虑数据点分布大多呈现不均匀性,LLE对近邻点的选取方式将会导致大量的信息丢失。根据其不足,提出一种基于数据点松紧度的局部线性嵌入改进算法——tLLE算法,针对数据点分布不均匀的数据集,tLLE算法能有效地进行维数约简,且具有比LLE更好的降维效果。在人造数据和现实数据上的嵌入以及分类识别结果表明了tLLE算法的有效性。     

基于分形变换对比度因子的图像数字水印新算法

分形变换的独特原理使得必须研究针对其特点的水印算法,通过对图像分形变换压缩过程中有关参数的分析和试验,选择变换过程中值域块和其最佳匹配块之间映射关系的对比度因子参数进行调制实现水印的嵌入和提取,该算法能够在低计算复杂度下嵌入较大的水印容量而不会对图像质量产生可感知的影响,提取算法避免了水印信息不同步造成的信息丢失问题,提高了水印检测的准确度。实验结果证明了该算法的有效性。