无线传感器网络中一种启发式最小连通支配集算法

针对最小连通支配集问题设计了一种具有较高能量效率的启发式算法．算法首先把网络中所有的节点作为最小连通支配集的一个初始解,然后利用启发式修剪策略剔除冗余节点从而减小最小连通支配集的大小,直到没有冗余节点存在．文中将算法分成集中式和分布式两种情况进行了详细讨论．仿真结果表明,由于实现简便,该算法与其他已有算法相比较,在算法复杂性和算法稳定运行时间上有一定的优势．     

同形映射下的曲面场景的匹配算法的性能分析

针对移动机器人在曲面场景的匹配问题中,同形约束用于解决极约束产生的匹配模糊性问题和发现新的匹配点.实际上是平面块对曲面进行近似逼近的过程.逼近程度和逼近性能需要有指标进行定性和定量的衡量.故提出了两种性能评价指标:平均映射误差和平均映射匹配对.仿真实验结果的分析证明,场景深度变化或者场景距离摄像机的距离变化,对立体匹配算法性能本身不受影响,但映射和建立匹配关系时所需要的同形矩阵的数量不同.而且,随着特征点的稠密度提高,曲面场景的稳定性降低.可随着迭代过程的进行,算法本身结果还趋于稳定.     

从工艺美术教育到艺术设计教育——称谓变迁背后的问题

本文通过分析工艺美术和艺术设计所依附的社会形态的差异性,以及对工艺美术教育和艺术设计教育称谓变迁现象的考察,显现当代中国工艺美术与艺术设计教育发展的脉络,并以此为基础,揭示当代艺术教育史的话语中,艺术设计与工艺美术之间关系复杂难言的根本原因。     

仿盲人摸巷机理的机器人导航方法分析

根据“盲人摸巷”及昆虫“用须探物”的启发,结合物体接触后能产生力的特性,提出基于接触交互信息的机器人导航方法,移动机器人与环境的接触力感觉来自移动机器人的触须,该触须可以是2个多自由度机械臂,其末端装有多维力传感器,或由弹性材料特制而成,达到完成探测、自我定位及局部路径规划任务的目标,是应用图像、光、电磁、声等原理的现有导航方法的很好补充。     

话题发现与追踪技术研究

话题发现与追踪以新闻流为处理对象,采用基于事件的信息组织方式进行研究,一直是自然语言处理领域里的热点。该研究借鉴大量相关研究尤其是信息检索中的经典模型和方法,取得了很大成功。首先介绍了话题发现与追踪的主要研究内容、评价方法以及发展历史;然后对其多个研究内容提出一个统一研究框架,并对该框架中的关键技术进行了详细分析;最后指出该领域中的关键问题及难点,并对未来研究做出展望。     

Mitigating Surface Deficiencies of Perovskite Single Crystals Enables Efficient Solar Cells with Enhanced Moisture and Reverse-Bias Stability

Metal halide perovskite single crystals are promising for diverse optoelectronic applications due to their outstanding properties. In comparison to the bulk, the crystal surface suffers from high defect density and is moisture sensitive; however, surface modification strategies of perovskite single crystals are relatively deficient. Herein, solar cells based on methylammonium lead triiodide (MAPbI₃) thin single crystals are selected as a prototype to improve single-crystal perovskite devices by surface modification. The surface trap passivation and protection against moisture of MAPbI₃ thin single crystals are achieved by one bifunctional molecule 3-mercaptopropyl(dimethoxy)methylsilane (MDMS). The sulfur atom of MDMS can coordinate with bare Pb²⁺ of MAPbI₃ single crystals to reduce surface defect density and nonradiative recombination. As a result, the modified devices show a remarkable efficiency of 22.2%, which is the highest value for single-crystal MAPbI₃ solar cells. Moreover, MDMS modification mitigates surface ion migration, leading to enhanced reverse-bias stability. Finally, the cross-link of silane molecules forms a protective layer on the crystal surface, which results in enhanced moisture stability of both materials and devices. This work provides an effective way for surface modification of perovskite single crystals, which is important for improving the performance of single-crystal perovskite solar cells, photodetectors, X-ray detectors, etc.     

RTDMiner: 基于数据挖掘的引用计数更新缺陷检测方法

在Linux内核等大型底层系统中广泛采用引用计数来管理共享资源.引用计数需要与引用资源的对象个数保持一致,否则可能导致不恰当引用计数更新缺陷,使得资源永远无法释放或者被提前释放.为检测不恰当引用计数更新缺陷,现有静态检测方法通常需要知道哪些函数增加引用计数,哪些函数减少引用计数.而手动获取这些关于引用计数的先验知识过于费时且可能有遗漏.基于挖掘的缺陷检测方法虽然可以减少对先验知识的依赖,但难以有效检测像不恰当引用计数更新缺陷这类路径敏感的缺陷.为此,提出一个将数据挖掘技术和静态分析技术深度融合的不恰当引用计数更新缺陷检测方法 RTDMiner.首先,根据引用计数的通用规律,利用数据挖掘技术从大规模代码中自动识别增加或减少引用计数的函数.然后,采用路径敏感的静态分析方法检测增加了引用计数但没有减少引用计数的缺陷路径.为了降低误报,在检测阶段再次利用数据挖掘技术来识别例外模式.在Linux内核上的实验结果表明,所提方法能够以将近90%的准确率自动识别增加或减少引用计数的函数.而且RTDMiner检测到的排行靠前的50个疑似缺陷中已经有24个被内核维护人员确认为真实缺陷.     

Base on the Public Scientific Quality Improvement Research on Risk Early Warning of Online Shopping

LTE-A network offers data rates up to 1 Gbps which is 10?×?faster than LTE catering to growing demand of users. LTE improves user experience by reducing latency and increasing bandwidth efficiency. The emerging services and key enhancements such as Further Enhancement of Downlink Multiple-Input Multiple-Output (MIMO), Heterogeneous Networks, and Carrier Aggregation (CA) in LTE-A has improved performance of LTE-A networks. Scheduling optimization still remains one of the biggest challenges in high speed data transmission network. Scheduling in LTE-A networks are performed at various levels; User Equipment (UE), Serving Gateway (SGW), Air Interface and eNodeB. Remote Radio Head (RRH) is an extremely specialized device installed at antenna of eNodeB for optical to electrical signal conversion, amplification of signals and Uplink and Downlink Scheduling. Resource scheduling at Antenna of eNodeB module is constituted as a significant research optimization area. This paper proposes a soft computing based scheduler for RRH. Results of proposed technique are evaluated on Fairness Index, Throughput, Spectral Efficiency and Rank Indicator Distribution. The proposed algorithm aims to improve performance of scheduling. From experimental results, it is observed that proposed model succeeds to achieve significantly better performance as compared to state-of-art algorithms.

     

双向特征融合的快速精确任意形状文本检测

现有的基于分割的场景文本检测方法仍较难区分相邻文本区域,同时网络得到分割图后后处理阶段步骤复杂导致模型检测效率较低.为了解决此问题,该文提出一种新颖的基于全卷积网络的场景文本检测模型.首先,该文构造特征提取器对输入图像提取多尺度特征图.其次,使用双向特征融合模块融合两个平行分支特征的语义信息并促进两个分支共同优化.之后,该文通过并行地预测缩小的文本区域图和完整的文本区域图来有效地区分相邻文本.其中前者可以保证不同的文本实例之间具有区分性,而后者能有效地指导网络优化.最后,为了提升文本检测的速度,该文提出一个快速且有效的后处理算法来生成文本边界框.实验结果表明:在相关数据集上,该文所提出的方法均实现了最好的效果,且比目前最好的方法在F-measure指标上最多提升了1.0％,并且可以实现将近实时的速度,充分证明了该方法的有效性和高效性.