The varied foraging mode of the subtropical eublepharid gecko Goniurosaurus kuroiwae orientalis

Insectivorous lizards usually employ one of two foraging modes, ambush ‘sit-and-wait’ or active ‘widely foraging’, but in the Gekkonomorpha the situation has been unclear. Therefore we quantified the foraging mode of the eublepharid gecko Goniurosaurus kuroiwae orientalis on Tokashikijima near Okinawajima, Japan, in September 1999. The taxon is rare, endangered and protected. On the selected semi-urban study site the gecko was confirmed as a nocturnal cursorial gecko, which also ascends trees up to 1.8 m. Foraging regimes of marked individuals, during repeated 30-min observation periods, could vary from immobility throughout, to active locomotion 84% of the time. Foraging mode was not significantly affected by sex, body length, mass or state of tail regeneration of the individuals, or by the timing of the observation. In males, at higher temperatures single moves became longer, while time allocated to moving was stable. In both sexes longer observation bouts included longer sitting pauses, which, given stable move duration, reduced the share of time spent moving and the moves per minute. Goniurosaurus kuroiwae orientalis averaged moving during 23% of the time, moving 0.39 times per minute, the moves lasting 77 s and the pauses lasting 745 s. Compared to other geckos, G. k. orientalis may be defined as a widely foraging animal, despite its mixed behaviour. This result supports the definition of the Eublepharidae as widely foraging and is compatible with two hypotheses, that early Gekkonomorpha were widely foraging or that they had an intermediate, undefined, foraging mode.     

基于相敏检波电路的功率因数测量

电网通过电流及电压互感器后,利用相敏检波及相位补偿电路,实现功率因数实时、准确的测量.为电力管理系统提高电网功率因数提供借鉴.     

高职院校食堂自主管理中HACCP体系及其实施

餐饮业作为食品经营行业之一,其安全问题受到了越来越多的关注。高职院校师生用餐人数多,分布相对集中,更易发生集体性食物中毒,加强管理显得尤为重要。高职院校食堂必须改变传统的管理模式,采用现代的、科学的管理理念和管理方法。本研究通过校园自主管理HACCP体系的培训和实施,分别在培训前和培训后,以现场稽查和微生物检测的方式,进行培训实施后的成效评估。     

基于模糊熵和结构特征的边缘检测方法

为了使检测的图像边缘结核定位好,并且产生边续的精细边缘,同时能有效滤除边缘图像中的噪声,提出一种基于模糊熵与边缘结构特征相结合的新的图像边缘检测方法。该方法首先将图像的灰度值特征空间转换为模糊熵特征空间,计算出每个像素在模糊熵特征空间的相异测度,结合3×3 邻域内的12种有效的边缘结构,提取图像中每个像素的结构信息测度图像阵列和方向信息测度图像阵列,然后对结构信息测度图像阵列和方向信息测度图像阵列实施非极大抑制,确定最终的边缘像素图像。实验结果表明该方法所检测出的图像边缘细节丰富,单像素宽,定位准确,有较好的抗噪性。     

虚拟数控车削系统中碰撞检测问题的研究

碰撞检测问题是虚拟数控车削系统的不可或缺的部分,有效、精确的碰撞检测对提高数控环境的真实性、增强虚拟环境的沉浸感有着非常重要的意义．碰撞检测问题就是测试物体之间是否存在相交的问题．采用合理的空间分割系统和层次包围盒是减少所需测试的多边形数目、提高碰撞检测的速度的有效方法．实验证明,采用基于空间分割方法的八叉树算法有效地提高了检测的速度,保证了虚拟动态加工环境中碰撞检测的实时性．     

基于近岸系列分层竖井的海底地下水排泄及其营养盐输入研究

以广西北海市近岸系列分层竖井为研究对象,通过现场地下水监测和采样分析,结果显示竖井含水层整体上为靠近海边咸化明显,其中I承压水咸化程度较大,咸化程度差异主要受海水入侵及海水高位养殖影响。地下水SiO₃-Si由于受"盐效应"影响导致其自内陆向海岸浓度升高,而低含量的PO₄-P含量可能是铁氧化物的吸附清除作用引起的;潜水层中的NH₄-N和NO_x-N含量整体高于承压水的,高位养殖池塘水的下渗造成NH₄-N在近海岸的潜水层含量最高,而NO_x-N受反硝化过程影响,导致自内陆至海岸明显减小。研究区地下水氮磷比(N/P)平均值为595.85,高N/P比值地下水排泄进入海洋,将促进浮游生物营养盐结构由N限制转向P限制。     

一种面向自动驾驶路况的目标检测算法

为了对常见的行人和车辆进行检测,采用自行标注的数据集,通过基于faster region-based convolutional neural network （RCNN）框架的算法进行调参与优化. 主干网络采用轻量化网络MobileNetv2,在原生锚框的基础上,区域建议网络（RPN）部分增加2个面积尺度,检测部分使用感兴趣区域（ROI）Align结构,减少特征图映射和均分过程中的误差. 实验结果表明：使用faster RCNN目标检测网络,可以有效完成行人和车辆的检测任务,整体效果良好.     

基于k循环随机序列的动态缓冲区溢出防御

面向Intel 80×86体系结构和C/C++语言,介绍了栈缓冲区溢出攻击的基本原理及攻击模式,分析了现有的动态防御典型方案的优点与不足.结合基于随机地址空间与签名完整性的防御思想,提出了一种基于k循环随机序列的动态缓冲区溢出防御方案,该方案能够在极大概率下防御多种模式的缓冲区溢出攻击,解决了"连续猜测攻击"的问题,并使软件具有一定的容侵能力.     

Harris角点检测算法的优化研究

在研究Harris角点检测算法时,发现检测出的角点常常会受到噪声的影响.通过优化Har-ris角点响应函数,避免了角点响应函数中k值的影响,提高了每个目标像素点的响应值精度.把该方法运用到人脸特征的检测中,实验结果表明,该算法和传统的Harris角点检测算法相比抗噪能力有较大提高.     

A Fault Detection Mechanism in Erasure-Code Byzantine Fault-Tolerance Quorum

Fault-tolerance is increasingly significant for large-scale storage systems in which Byzantine failure of storage nodes may happen. Traditional Byzantine Quorum systems that tolerate Byzantine failures by using replication have two main limitations： low space-efficiency and static quorum variables. We propose an Erasure-code Byzantine Fault-tolerance Quorum that can provide high reliability with far lower storage overhead than replication by adopting erasure code as redundancy scheme. Through read/write operations of clients and diagnose operation of supervisor, our Quorum system can detect Byzantine nodes, and dynamically adjust system size and fault threshold. Simulation results show that our method improves performance for the Quorum with relatively small quorums.