一种基于数据竞争的高分辨率图像的聚类分割算法

近些年,机器人技术得到了迅猛的发展,应用越来越广泛.随着机器人技术的推广和普及,对机器人使用的要求也越来越高,其中对智能机器人的要求尤显迫切.机器视觉是智能机器人研究领域的一个重要研究方向.在机器人视觉系统中,核心问题是目标提取,对目标实时、准确、快速提取的关键技术是图像分割.由于机器人感知的环境的复杂性及目标的多样性,往往导致机器人感知获得的图像数据量较大且图像本身存在不可预知的复杂性,这就对准确的目标分割和提取处理提出了挑战性问题.本文针对高分辨率图像数据集的分割处理,提出一种新的聚类算法,即根据数据点能量和的大小识别类代表点和类成员点,通过数据点间的竞争识别出最有能力成为簇成员的数据点,并将其与mean shift聚类算法有效地结合应用于彩色图像分割问题中,能够快速高效地实现高分辨率图像的目标分割,并得到较好的图像分割效果.实验结果表明,本文算法在分割效果和分割效率上明显优于传统聚类算法.     

A low power and small area all digital delay-locked loop based on ring oscillator architecture

A 133-500 MHz,5.2 mW @500 MHz,0.021 mm2 all digital delay-locked loop(ADDLL)is presented.The power and area reduction of the proposed ADDLL are achieved by implementing a high frequency ring oscillator(ROSC)to count the reference clocks such that the one-clock cycle delay chain and the phase detector in a conventional Master block are no longer needed.The proposed ADDLL has better immunity to PVT(process,voltage,and temperature)than most conventional DLLs,which do not update the control word signals after the locking process,since the control signals for slave delay line are updated in every 256 reference cycles.Fabricated in 0.13 um CMOS process,the measured RMS jitter is 10.83 ps at 500 MHz while the RMS jitter of the input signal is 9.97 ps.     

基于兴趣边缘优化的壁画影像与激光扫描数据非刚性配准

将壁画影像与激光扫描数据配准,并进行定位和纠正在壁画的数字化保护中有非常重要的意义.本文以激光扫描数据强度信息为中介,提出了一种基于兴趣边缘优化的壁画影像与激光扫描数据的非刚性配准方法:由激光扫描数据生成强度影像,以壁画彩色影像的兴趣边缘和强度影像的梯度场作为配准基元,在影像刚性配准基础上,对每条兴趣边缘进行优化配准,然后以优化后边缘的特征点为控制点,构造影像之间的非刚性变换模型,完成壁画影像与激光扫描数据的配准.实验结果表明本方法在不同数据中都能获得较高的配准精度.     

人体步行规律与仿人机器人步态规划

从仿生学角度分析了人体的步行运动规律,提出了一种基于人体运动规律的仿人机器人步态参数设定方法.首先对人体步行运动数据进行捕捉并分析,得出人体各步态参数间的函数关系,以人体步行相似性作为评价指标,提出仿人机器人步态参数的设定方法.其次,通过分析人体在步行过程中的补偿支撑脚偏航力矩的基本原理,提出了基于双臂及腰关节协调运动的仿人机器人偏航力矩补偿算法,以提高仿人机器人行走的稳定性.最后通过仿真及实验验证了所提出的步态规划方法的正确性及有效性.     

IT投资风险预警的研究

进行IT投资风险预警的基础是对IT投资风险域进行分析;文章提出了基于IT投资目标、信息化生命周期以及企业架构的IT投资风险域的三维框架,并引入贝叶斯网络的原理建立预警机制。最后,在此基础上进行了IT投资风险预警的案例分析。     

Negative effects of sufficiently small initial weights on back-propagation neural networks

In the training of feedforward neural networks,it is usually suggested that the initial weights should be small in magnitude in order to prevent premature saturation.The aim of this paper is to point out the other side of the story:In some cases,the gradient of the error functions is zero not only for infinitely large weights but also for zero weights.Slow convergence in the beginning of the training procedure is often the result of sufficiently small initial weights.Therefore,we suggest that,in these cases,the initial values of the weights should be neither too large,nor too small.For instance,a typical range of choices of the initial weights might be something like(0.4,0.1) ∪(0.1,0.4),rather than(0.1,0.1) as suggested by the usual strategy.Our theory that medium size weights should be used has also been extended to a few commonly used transfer functions and error functions.Numerical experiments are carried out to support our theoretical findings.