Parallel vector processing of multidimensional orthogonal transforms for digital signal processing applications

This paper presents vector and parallel algorithms and implementations of one- and two-dimensional orthogonal transforms. The speed performances are evaluated on Cray X-MP/48 vector computer. The sinusoidal orthogonal transforms are computed using fast real Fourier transform (FFT) kernel. The non-sinusoidal orthogonal transform algorithms are derived by using direct factorizations of transform matrices. Concurrent processing is achieved by using the multitasking capability of Cray X-MP/48 to transform long data vectors and two-dimensional data vectors. The discrete orthogonal transforms discussed in this paper include: Fourier transform (DFT), cosine transform (DCT), sine transform (DST), Hartley transform (DHT), Walsh transform (DWHT) and Hadamard transform (DHDT). The factors affecting the speedup of vector and parallel processing of these transforms are considered. The vectorization techniques are illustrated by an FFT example.This work is supported in part by the National Science Foundation, Pittsburgh Supercomputing Center (grant number ECS-880012P) and by the PEW Science Education Program.     

用整数Haar小波变换和分块DPCM实现静止图像数据的信息熵保持型压缩编码

本文首先讨论了整数Ｈａａｒ小波的构造方法,然后利用分块ＤＰＣＭ与整数Ｈａａｒ小波变换进行信息保持图像数据压缩,该方法可进行实时处理,硬件实现简单,可并行处理,实验结果表明,该方法是一种有效的图像数据压缩方法。     

Tree structures for orthogonal transforms and application to the Hadamard transform

Tree structures for computing orthogonal transforms are introduced. Two cases, delay trees and decimation trees, are investigated. A simple condition, namely the orthogonality of branches with a common root, is shown to be necessary and sufficient for the overall transform to be orthogonal. Main advantages are structural simplicity and a number of operations proportional to N Log₂N.Application of the tree structures to the Walsh-Hadamard Transform (in natural, sequency and dyadic order) is presented. A single module can be multiplexed or used in parallel in order to perform all operations. Such a system is shown to be well suited for hardware implementation.     

Hadamard变换和Haar变换谱系数的图形转换法

本文根据Hadamard变换谱系数（had谱系数）与布尔列矢量的关系以及Haar为换谱系数（ha谱系数）与布尔列矢量的关系,提出了had谱系数与ha谱系数之间的图形转换方法,并举例说明转换过程,该方法揭示了系数图与ha系数图之间的内在联系,具有简单,直观的特点。     

A commutative digital image watermarking and encryption method in the tree structured Haar transform domain

In this paper a commutative watermarking and ciphering scheme for digital images is presented. The commutative property of the proposed method allows to cipher a watermarked image without interfering with the embedded signal or to watermark an encrypted image still allowing a perfect deciphering. Both operations are performed on a parametric transform domain: the Tree Structured Haar transform. The key dependence of the adopted transform domain increases the security of the overall system. In fact, without the knowledge of the generating key it is not possible to extract any useful information from the ciphered-watermarked image. Experimental results show the effectiveness of the proposed scheme.     

一种快速有效的并行条纹细化方法

本文研究的是快速有效的并行细化方法。通过巧妙地构造细化处理数据的结构和程序的优化设计，大大地提高了细化算法的运算速度。同时也保证了细化后条纹的连续性、拓扑性不变，确保细化纹线的定中性，另外此细化算法具有一般性，对其他具有拓扑特征的二值图象也可进行细化处理。     

改进的Hough变换及其并行计算

本文针对传统的Ｈｏｕｇｈ变换不能测出直线的连贯性、无法求得线段的长度的缺点，提出一种改进的Ｈｏｕｇｈ变换算法，使检测结果为线段的起始及长度。     

基于散列技术的并行流水线Join算法的设计与评价

本文研究了一种新型的混合式通用并行结构上基于散列技术的并行连接算法的实现技术，并建立了一种性能分析模型，这种模型特别刻划了系统中三种主要资源ＣＰＵ、磁盘、互联网之间的操作重叠，我们以这一模型为基础深入分析了并行Ｈｙｂｒｉｄ散列连接算法（称为ＰＨＨＪ）和它的一种基于小关系复制技术的变形算法（称为ＰＨＨＪＲ）的性能及与系统结构之间的关系。我们还结合传统的位向量过滤技术提出一种用对称位向量和内外关系角色     

基于球面Haar小波和卷积神经网络的飞行员虹膜识别

虹膜识别面临两个重要的问题：一是如何精细分解与重构虹膜球面图像;二是如何识别虹膜图特征。虹膜表面几何位置信息是一种重要的信号,传统的虹膜识别通常使用虹膜图像的平面特征,然而人的眼睛是一种球体,从平面图像难以提取到虹膜球体的几何特征。针对平面特征容易出现虹膜纹理的扭曲和失真等问题,该文建议一种正交对称的球面Haar小波(OSSHW)基,对球面虹膜信号进行多尺度分解与重构,获得更精细的虹膜曲面几何特征,同时对比球谐函数和半正交或正交球面Haar小波基的虹膜球面信号特征提取能力。在此基础上,该文提出一种基于卷积神经网络(CNN)和正交对称的球面Haar小波的虹膜识别方法,它能够有效捕获虹膜球体曲面的局部精细特征,比半正交或正交球面Haar小波基具有更强的虹膜识别能力。     

基于树结构矢量分类的小波图像图编码矢量量化

本文基于零树编码,矢量分类和网络编码量化的思想,提出了对小波图象采用树结构矢量组合和分类后进行网络编码矢量量化的新方法,该方法充分利用了带系统的带间和带内的相关性,分类信息上中用比特数少,对重要类矢量实行加权网络编码矢量量化,利用卷积编码扩展信号空间以增大量化信号间的欧氏距离,用维特比算法搜索最优量化序列,并采用基于人眼视觉性特性的加权均方误差准则作为失真度量和码字匹配,提高了量化增益,仿真结果表明,该方法编码计算复杂度适中,解码简单,可达到很好的压缩效果。     

Some properties and fast algorithms of slant transform in image processing

A first order stationary Markov process model has been considered for image processing problems. A relative performance measure of unitary transforms to image data has been defined. It has been proved that the slant transform is superior to Walsh-Hadamard transform in this relative performance measure for positive correlation under the assumed model. A lower bound of relative performance has also been found. Furthermore, fast algorithms for computing diagonal elements of any slant transformed matrix have been presented. Finally, it has been shown how slant transform can be modified to improve the relative performance.     

Parallel GMRES solver for fast analysis of large linear dynamic systems on GPU platforms

In this paper, we propose an efficient parallel dynamic linear solver, called GPU-GMRES, for transient analysis of large linear dynamic systems such as large power grid networks. The new method is based on the preconditioned generalized minimum residual (GMRES) iterative method implemented on heterogeneous CPU–GPU platforms. The new solver is very robust and can be applied to power grids with different structures as well as for general analysis problems for large linear dynamic systems with asymmetric matrices. The proposed GPU-GMRES solver adopts the very general and robust incomplete LU based preconditioner. We show that by properly selecting the right amount of fill-ins in the incomplete LU factors, a good trade-off between GPU efficiency and convergence rate can be achieved for the best overall performance. Such tunable feature can make this algorithm very adaptive to different problems. GPU-GMRES solver properly partitions the major computing tasks in GMRES solver to minimize the data traffic between CPU and GPUs to enhance performance of the proposed method. Furthermore, we propose a new fast parallel sparse matrix–vector (SpMV) multiplication algorithm to further accelerate the GPU-GMRES solver. The new algorithm, called segSpMV, can enjoy full coalesced memory access compared to existing approaches. To further improve the scalability and efficiency, segSpMV method is further extended to multi-GPU platforms, which leads to more scalable and faster multi-GPU GMRES solver. Experimental results on the set of the published IBM benchmark circuits and mesh-structured power grid networks show that the GPU-GMRES solver can deliver order of magnitudes speedup over the direct LU solver, UMFPACK. The resulting multi-GPU-GMRES can also deliver 3–12× speedup over the CPU implementation of the same GMRES method on transient analysis.     

并行测试的测试调度及其控制

本文首先从理上比较严格的证明了测试调度问题是多项式可解的，提出了一种Ｏ（ｎ＾２）的最优测试调度策略。最后，阐述了一种测试调度的控制策略，使得多路转换器控制输入最少。     

Encoding true-color images with a limited palette via soft vector clustering as an instance of dithering multidimensional signals

One of the classic problems of digital image processing is to encode true-color images for the optimal viewing on displays with a limited set of colors. A major manifestation of optimal viewing in this regard is to maximally remove parasitic artifacts in the degraded encoded images such as the contouring effect. Several robust attempts have been made to solve this problem over the past 50 years, and the first contribution of this paper is to introduce a simple – yet effective – novel solution that is based on soft vector clustering.The other contribution of this paper is to propose the application of the soft clustering methodology deployed in our color-encoding solution for the dithering of multidimensional signals. Dithering essentially adds controlled noise to the analog signal upon its digitization so that the resulting quantization noise is dispersed over a much wider band of the frequency domain and is therefore less perceptible in the digitized signal. This comes of course at the price of more overall quantization noise. Dithering is a vital operation that is performed via well-known simple schemes upon the analog-to-digital conversion of one-dimensional signals; however, the published literature is still missing a general neat scheme for the dithering of multidimensional signals that is able to handle arbitrary dimensionality, arbitrary number and distribution of quantization centroids, and with computable and controllable noise power. This gap is also filled by this paper.     

基于矢量分解和相位截取Gyrator变换的图像加密

为了提高光学图像加密系统的安全性,采用了矢量分解和相位截取Gyrator变换进行图像加密。原始图像和随机相位函数叠加后做Gyrator变换,矢量分解将Gyrator变换域信息分解为一个随机相位函数和一个复函数。随机相位函数和复函数分别做不同变换角度的Gyrator变换后截取相位得到一个公钥和加密图像,截取振幅得到两个非对称相位密钥。进行了理论分析和数值验证,同时,设计了解密光学装置。结果表示:两个非对称相位密钥做为私钥,三个Gyrator变换角度做为额外密钥,这对增强系统的安全性是有帮助的。     

曲波变换和最小二乘支持向量机的图像压缩算法

为了提高图像压缩质量,针对传统压缩算法的不足,提出一种曲波变换和最小二乘支持向量机相融合的图像压缩算法。首先采用曲波变换把图像分解为不同尺度和不同方向的曲波系数,并采用熵编码对粗尺度层曲波系数进行压缩,然后利用最小二乘支持向量机对细尺度层中不同方向的曲波系数进行学习,并通过和声搜索算法优化最小二乘支持向量机,实现细尺度层曲波数的压缩,最后采用图像压缩仿真实验测试其性能。结果表明,曲波变换和最小二乘支持向量机相融合的图像压缩算法提高了图像压缩的峰值信噪比,加快了图像压缩的速度,获得了更好的图像压缩效果。     

图形测试:多工位模拟和混合信号器件并行测试效率的关键

多工位测试是许多模拟和混合信号器件生产厂家大批量测试的基石.一直以来,模拟和混合信号器件测试系统面临架构的不足致使并行测试效率(PTE:ParallelTestEfficiency)降低。测试系统架构不断地发展寻求并行测试效率的提高。这一组文章说明并讨论多种测试系统硬件和软件的设计改进,产生更高的并行测试效率。     

图形测试:多工位模拟和混合信号器件并行测试效率的关键(第二部件)

多工位测试是许多模拟和混合信号器件生产厂家大批量生产测试的基石.随着测试工位数的增加,模拟和混合信号测试系统设计者需要努力克服历来限制并行测试效率(PTE:Parallel Test Efficiency)的架构上的不足。在第一部分中,我们介绍了并行测试效率PTE及其随工位数增加对测试成本的影响,突出了一些效率高的多工位测试系统的重要特性。第二部分我们将介绍pattern-based测试和有关的SmartPin硬件和软件。     

Bi_(12)SiO_(24)晶体用于光并行逻辑运算时输出质量及写入灵敏度的分析

以 Bi_(12)SiO_(24)晶体制成的实时器件 BSO-PROM 可用于光并行逻辑运算。本文分析了影响其输出对比度、实际分辨率及写入灵敏度的因素;并给出了获得高对此度、高灵敏度应满足的条件。