基于SIFT 特征的目标多自由度mean-shift 跟踪算法

传统的mean-shift跟踪算法不能跟踪目标的旋转、缩放运动,且常常因此造成定位不准.鉴于此,将尺度不变特征变换（SIFT）特征检测融入到mean-shift跟踪过程,提出SIFT特征点的尺度变化与目标的尺度变化成正比,特征点主方向变化与目标旋转角度一致,给出了基于SIFT特征的自适应目标尺度、方向计算方法,且利用带方向、可变带宽的椭圆核改进传统的mean-shift跟踪方法.实验表明,该算法能够较好地跟踪目标的旋转、缩放运动,定位也更准确.     

尺度与颜色不变性图像特征描述

尺度不变特征变换是目前公认的鲁棒性最强的图像特征描述方法之一,在尺度不变性和几何不变性方面具有较好的特性,但该方法主要适用于灰度图像,对图像颜色的区分能力不强,因此,一些对象可能会因为颜色的不同而被错误的区分.另外,尺度不变特征变换对关键点局部范围内描述子主方向的依赖性非常强,直接决定了匹配的正确率,但是研究表明,主方向分配产生的误差仅有三分之二左右能控制在[-20。,+20。]范围内,因此部分特征会有三分之一的概率因为主方向分配的误差较大而不能正确匹配.针对以上两个问题,本文提出了一种具有颜色和尺度不变性的局部特征描述方法,颜色不变性通过将RGB图像转换到高斯颜色模型下实现,特征描述过程中不再分配主方向,而用局部相对方向,尺度不变性通过构建高斯金子塔实现.实验选取阿姆斯特丹数据集图像进行了测试,结果表明本文方法比传统尺度不变特征变换方法,在特征点的数目、分布均匀性以及匹配精度方面均有所提高.     

基于改进几何哈希法的矢量图形检索

针对图形文件以矢量方式记录几何实体的特点,提出了一种基于改进几何哈希法的图形检索算法。与图像的对应处理方式相比,该算法充分利用图形的矢量信息,以实体基准点和实体的准确几何形式描述几何实体;在检索过程中,每个图形实体仅有基准点参与排序和查询,哈希表数据组织形式简单,查询速度快;通过验证图形实体属性,保证了查询结果的准确性。最后以矢量图形文件为图例,验证了方法的有效性。     

Fourier methods for quasi‐periodic oscillations

Quasi‐periodic oscillations and invariant tori play an important role in the study of forced or coupled oscillators. This paper presents two new numerical methods for the investigation of quasi‐periodic oscillations. Both algorithms can be regarded as generalizations of the averaging and the harmonic (spectral) balance methods. The algorithms are easy to implement and require only minimal a priori knowledge of the system. Most importantly, the methods do not depend on an a priori co‐ordinate transformation. The methods are applied to a number of illustrative examples from non‐linear electrical engineering and the results show that the methods are efficient and reliable. In addition, these examples show that the presented algorithms can also continue through regions of sub‐harmonic (phase‐locked) resonance even though they are designed only for the quasi‐periodic case. Copyright © 2006 John Wiley & Sons, Ltd.     

一种基于R树的图像检索方法

把图像的七个不变矩组合成一个七维向量,在R-树索引结构的基础上设计和实现了此向量的多维索引结构及相关操作。试验结果显示使用该索引方法大大提高了图像检索的效率。     

Static Output Feedback Stabilization: An ILMI Approach

In this note, the static output feedback stabilization problem is addressed using the linear matrix inequality technique. A necessary and sufficient condition for static output feedback stabilizability for linear time-invariant systems is derived in the form of a matrix inequality. The extension of the result to H_∞ control is studied. An iterative LMI (ILMI) algorithm is proposed to compute the feedback gain. Numerical examples are employed to demonstrate the effectiveness and the convergence of the algorithm.     

C上全纯自映射的完全不变域

设ｆ（ｚ）是Ｃ上全纯自映射，０和是它的本性奇点，其中本文讨论ｆ（ｚ）的完全不变域的性质，证明了如果Ｄ是ｆ（ｚ）的完全不变域，那么Ｄ一定是二连通的且Ｄ包含了的奇异值集；此外，如果ｆ（ｚ）是有限型的，则Ｄ＝Ｆ（ｆ）．由此结论可以推出ｆ（ｚ）最多只有一个完全不变域在Ｆ（ｆ）中     

执行器饱和单输入线性系统的分析和设计

若线性系统可被某状态线性反馈全局渐近镇定 ,饱和介入反馈后 (具有反馈的 )系统稳定性或得以保持 ,或降为区域渐近稳定 .本文作者就这一问题给出简洁判据 ,且对后者计算其不变吸引 (超 )椭球 ;在此基础上用叠代法求取使椭球尽量大的反馈增益 ;最后用算例说明方法的有效性 .     

A comparison between classical robust stability conditions

The paper is focused on the comparison between some classical robust stability conditions for continuous‐time linear time‐invariant systems. Such conditions are given as lemmas in the paper, since their statements include some generalizations, needed in order to better compare (and use) them. The analysis is carried out by comparing pairwise the families of systems whose stabilization through a given compensator is guaranteed by each of the considered robust stability conditions. Some properties of the families are formally derived and stated, and some very simple examples arc exhibited in order to illustrate the presented properties and comparisons. A theorem, representing an extension of one of the considered conditions, is proposed in order to overcome some difficulties arising in the use of the classical result. Copyright © 2004 John Wiley & Sons, Ltd.