图像纹理合成的研究应用

在此介绍了纹理和纹理合成的基本概念,给出了纹理合成常见的几种算法,通过分析块拼接纹理合成算法,对块拼接纹理合成算法进行改进,将块拼接的原理应用到图像拼接上,并对象素加入与人视觉有关的权值进行接缝处理。实验结果表明,该方法简单实用,对于基本的由不同方位拍摄的图像都可以通过该算法进行拼接,并通过对接缝处的处理,取得了较好的结果。     

一种改进的纹理合成算法

基于已有的纹理合成算法,提出将单象素点的纹理合成改进为矩形纹理区域的合成,以加速合成过程,匹配矩形纹理区域时采用提取纹理结构信息的方法来实现,取得了较好的合成结果.实验结果表明,改进的算法无论在合成速度和合成质量上都有较大的提高,对于大纹理的合成,速度更快,而且有广泛的适用性.     

一种快速块拼贴纹理合成算法

提出一种改进的基于样图的块拼贴纹理合成算法.针对拼贴纹理块时产生的局部性边界不匹配现象,用图的切割算法寻找块重叠区域的最优拼接路径,有效地提高了纹理合成质量.同时采用多分辨率图像金字塔构建输入纹理图像,并结合螺旋线搜索策略和加速算法提高纹理合成速度.     

一种基于圆形区块随机增长的多样图约束纹理合成算法

本文提出了一种基于圆形区块随机增长的多样图纹理合成算法改善了扫描线算法所带来的锯齿效应。算法利用高斯分布约束各个输入样图在合成结果中的比例,采用基于梯度引导的泊松平滑处理相邻纹理块之间的过渡区域,并在合成匹配函数中引入结构特征约束。实验结果表明,算法在引入比例约束的同时有效地保持了视觉效果和纹理结构的连续性。     

基于区域纹理合成的图像修补算法

针对传统修补技术在图像恢复中出现的纹理模糊现象,该文根据纹理合成技术,提出了一种区域纹理合成的修补算法。算法根据破损点邻域特征选择模板,去除模板内的破损点后进行匹配,在搜索范围内寻找最佳匹配点。该文采用由边缘向中心逐步推进的修补次序,提高破损区域内纹理结构延伸方向的正确性以及连续性,避免出现纹理模糊现象。算法能够适用于多种性质的纹理,在处理真实图像时使人在视觉上获得满意的效果。     

基于对抗生成网络的纹理合成方法

纹理合成是计算机图形学、计算机视觉和图像处理领域的研究热点之一。传统的纹理合成方法往往通过提取有效的特征样式或统计量并在该特征信息的约束下生成随机图像来实现。对抗生成网络作为一种较新的深度网络形式,通过生成器和判别器的对抗训练能够随机生成与观测数据具有相同分布的新数据。鉴于此,提出了一种基于对抗生成网络的纹理合成方法。该算法的优点是不需要经过多次迭代就能够生成更真实纹理图像,且生成图像在视觉上与观测纹理图像一致的同时具有一定随机性。一系列针对随机纹理和结构性纹理的合成实验验证了该算法的有效性。     

基于结构张量方向引导的改进的W-L纹理合成算法应用研究

文章设计的算法在Wei和Levoy的思想基础上首先采用结构张量计算出引导图像的方向场,其次在方向场的引导下对样本纹理进行旋转,最后采用邻域相似性查找方法进行纹理合成,为了提升合成效率,文章通过采用向量化树结构的量化方法进行加速,取得了良好的实验效果。文章设计的算法不仅可以应用于方向纹理合成领域,同时通过拓展相关技术也可运用到非真实感绘制与渲染方面。     

基于纹理平滑度的视点合成失真优化快速算法

针对3D-HEVC中深度图编码采用的视点合成失真优化方法的高复杂度问题,提出一种基于纹理平滑度的快速算法。首先结合帧内DC预测特性和统计学方法分析平坦纹理图中像素规律并设定基于纹理图平坦度的跳过准则;然后在深度图编码采用视点合成失真优化方法时提前分离出纹理图平坦区域所对应的深度图区域,并终止该区域像素基于虚拟视点合成的视点合成失真计算过程。实验结果证明该算法的有效性,能在保持编码质量的同时减少大量编码时间。     

基于图像修复技术的压缩方法的研究

通过对块状缺损图像修复技术的研究,提出一种改进的纹理合成修复算法。继而依据粗糙度的分析对图像块进行分类,根据修复算法特性人为丢弃部分图像块,并结合JPEG算法进行压缩编码。在相近的视觉效果下,可以使图像压缩率得到显著的提高。     

利用自适应算法的快速估计参数进行图像纹理分割

本文提出使用Ｗｉｄｒｏｗ－ＨｏｆｆＬＭＳ自适应算法对二维随机场ＳＡＲ模型参数进行快速估计并报导了使用ＳＡＲ模型的自适应估计参数进行有监督纹理分类及纹理分割的实验结果，实验结果表明，ＳＡＲ模型的自适应快速估计参数能够比较有效地进行纹理识别。     

基于图着色的加权优先D2D资源分配算法研究

针对蜂窝用户与D2D用户所构成的异构网络系统中同频干扰问题,提出一种基于图着色的加权优先D2D资源分配算法.该算法不仅允许多个D2D用户复用一个蜂窝用户资源,而且能够实现简单功控.首先建立异构干扰图,对系统终端用户及干扰类型进行分类异构.然后计算着色优先级,考虑各种影响因子以提升算法的实用性.最后再由分配结果进行组内功率控制,以满足绿色通信的要求.仿真表明,该算法不仅可以降低系统用户接入损失率,提高系统吞吐量,而且还减少了功率消耗.     

Radio interface protocols and radio resource management procedures for 5G new radio non-terrestrial networks

In Release 17, 3GPP introduced adaptations and enhancements to the 5G new radio (NR) specification to support non-terrestrial networks (NTNs) operation. The main challenges were due to long propagation delays, especially in GEO deployments, and the movement of the satellites in LEO deployments. In this paper, we give an overview of the protocol adaptations to support NTNs. The main user plane protocol adaptations include changes to random access and hybrid automatic repeat request to due long propagation delays. The control plane protocol adaptations include a variety of mobility related enhancements for user equipment.     

Video transcoding proxy for 3Gwireless mobile Internet access

Transcoding proxies for mobile World Wide Web access have been developed. In the near future, wideband mobile communication systems such as IMT-2000 will emerge, and these proxies will have to cope with a variety of media such as video. In this article we identify issues involved in viewing a video stream in a mobile computing environment; to handle these issues, we propose a video transcoding system and its control method. These mechanisms provide stable transmission of video data and comfortable video viewing on user clients by estimating the communication conditions, client device capabilities, and user preferences     

基于负荷均衡的WCDMA网络资源优化探讨

随着数据业务的爆发增长,WCDMA网络负荷不断攀升,出现了业务质量变差的现象.通过研究多种控制资源拥塞的优化策略,提出了基于系统内、系统间的负荷均衡思路和方案,并结合各种场景进行了实验验证.结论：提高了系统容量和业务QoS,改善了网络性能和用户感知.     

对等环境中访问控制的方法

所有连接到网络中的用户都可以共享网络资源,在这种情况下,对于各种网络应用的攻击也是随时存在的,所以需要更为复杂的安全服务。本文提出一个使用SPKI证书在对等环境中进行访问控制的方案,这个方案是通过给网络中想要共享资源的用户提供证书的方法来进行访问控制的,根据分配给用户的访问权限来限制信息提供者的资源的使用。     

Spracherkennung mit Hidden Control Neural Networks

Hidden control neural networks (HCN networks) are suitable for a variety of pattern recognition techniques. The speech recognizer described here is built for speaker-independent single-word recognition and is intended to implement user interfaces to control devices via simple word-commands. To evaluate the speech recognizer, it has been applied to minimum pairs. Within a minimum pair two words differ only in a single phoneme. It was achieved to increase the recognition rate while taking those periods of time especially into account, that are found to contain the relevant difference.     

Security in enterprise networking: A quick tour

Enterprise networks are complex environments that involve the interconnection of a wide variety of computer systems such as portable PCs and personal digital assistants (PDAs), desktop PCs and workstations, servers, and mainframes, with a wide variety of communication channels such as dial-in and mobile access via modems, local area networks (LANs), wide area networks (WANs), and the Internet. The authors provide an overview of the major areas in the security of enterprise networks to show the variety of issues and techniques developed to address them. Our focus is on the ideas behind these techniques, which can be combined in many ways to create solutions that apply to different situations. The following areas are covered: confidentiality, preventing the disclosure of transmitted data to unauthorized parties; integrity, detecting modification, insertion, deletion, or replay of transmitted data; data-origin authentication, demonstrating that the origin of transmitted data is as claimed; nonrepudiation, preventing either the sender or receiver in a communication from denying their participation; user authentication, demonstrating that the identity of a user or system is as claimed; and access control, guarding against unauthorized use of resources, including the use of resources in an improper manner. We also look at some of the considerations that come into play in designing security solutions for the enterprise networking environment     

基于能量优化的微博用户转发行为预测

微博用户转发行为预测是微博社交网络消息扩散模型构建的基础,在舆情监控、市场营销与政治选举等领域有着广泛的应用.为了提高用户转发行为预测的精度,本文在MRF（Markov Random Field）能量优化框架下综合分析了用户属性与微博内容特征、用户转发行为约束、群体转发先验等因素对用户转发行为的影响,并在逻辑回归模型的基础上构造了相应的能量函数对用户转发行为进行了全局性的预测.实验结果表明,微博用户转发行为不仅取决于用户属性、微博内容等特征,而且也受到用户转发行为约束、群体转发先验等因素不同程度的影响.相对于传统算法,本文算法可以更准确地对用户转发行为进行建模,因而可获得更好的预测结果.     

The Concept of a Personal Distributed Environment for Wireless Service Delivery

Traditionally wireless service delivery has been restricted in scope to voice and text messaging and has been targeted at a single user device. Increasing interest is now being shown in ubiquitous delivery of services, including many forms of information and entertainment services, over a variety of wireless delivery mechanisms and to a collection of devices controlled by an individual user. These devices may be both local and remote to the user but their combination, and the services provided, form the user’s personal distributed environment (PDE). This paper reviews the concepts associated with a PDE and highlights some of the technical problems which require to be addressed in order to make such an environment a self-organising and attractive proposition from the user viewpoint. The paper defines the elements of a personal distributed environment and concentrates on the issues of personal access management and personal service management. It illustrates typical signalling exchanges required for the control of devices within the PDE and stresses the way in which the PDE may be attached to the core network. The paper will also illustrate a typical scenario in which a PDE might be deployed and will highlight the very important issues of security which must be incorporated in the overall PDE management.     

智能电视终端的多屏互动设计与实现

在三网融合的大背景下,包括手机、平板、智能电视等在内的各类智能设备层出不穷。为提高这类智能设备之间良好的互操作性和用户操控体验,本文面向智能电视终端,设计开发多屏互动系统,实现互联网视频内容搜索、跨屏视频内容推送及智能电视终端操控。介绍了系统组成及基本的工作流程,并描述了技术原理及实现。研发成果已在智能电视终端产品中成功部署。