在算法上

3D control of micropositioning, the environment for human-computer interaction based on the micro stereo image displaying, and microinjection are three focuses of this dissertation.The imaging models of CMO micro stereoscope and G micro stereoscope used in microvision subsystem are deduced. According to the imaging models, 3D visual servoing model based on image for micropositioning is set up. The technology of tracking and matching on the feature of image, which are relevant in visual servoing, is studied. Two arithmetic for avoiding obstacles, rolling/surrounding and curved tracking based on the depth first search, are further discussed.

在此基础上，对三维微定位控制技术、基于显微立体图像显示的人机交互操纵环境和微量注射等三个内容进行了重点研究：为了解决3D微定位问题，推导了显微视觉分系统采用的CMO型和G型体视显微镜光学成像模型，在此基础上建立了基于图像的3D显微视觉伺服定位控制模型，对视觉伺服中涉及的实时跟踪与立体匹配技术进行了研究，并进一步探讨了卷包裹法和曲线跟踪两种避障算法。

It has been proven that the textural information can be used to classification by means of test.

本文首先分析了遥感影像分类方法和纹理提取方法的研究现状，接着对地统计学的基本原理进行了论述，在这一理论的基础上，对变差函数应用于遥感影像的纹理描述和提取进行了论证和实验说明；并就一些具体应用问题如计算窗口大小、计算方向和计算步长的选择等问题进行了讨论，提出了一种自适应的确定纹理计算窗口大小的方法；同时，在现有算法的基础上提出了一种新的纹理计算方法――加权变差函数法。

As for the theory of route traffic state estimation, we firstly analyzed the traffic stream characters on the routes, and then we selected the license auto-recognized system as the input of our study.

在研究方法和理论创建上，注重了与实际工程应用的结合，文中提出的理论模型、算法建立在数据分析的基础上，保证了理论的真实可用性。

In the electric motor soft start question, uses the crystal thyratron intelligence module which controls by the singlechip to change the electric motor the initial firing current, the realization soft start; In the protection question, uses the singlechip to monitor the electric motor the movement, to lacks the breakdown implementation to separate the electrical machinery movement immediately, according to certain algorithm judgment electric motor whether crosses the class, like tallies has flowed the condition is the separation electrical machinery movement.

在电动机软起动问题上，采用由单片机控制的晶闸管智能模块改变电动机的起动电流，实现软起动；在保护问题上，采用单片机监视电动机的运行，对缺相故障实行立即断开电机运行，根据一定算法判断电动机是否过流，如符合过流条件即断开电机运行。

The reduction polynomial of a trinomial is represented by the binary vector. On this base, we optimize the algorithm to improve the performance of critical module both on area and on operation time.

在此基础上，我们对关键运算模块的实现进行了算法级的优化，采用了—些有效的方法，使得在较小面积上得到了较快速度，提高了芯片的整体性能。

Firstly, we define the two """"views""""of examples by choosing two Classifiers based on different algorithm theory. And then we proposed a example selection method based on the consistence, using two classifiers: Transductive HMM and fhTBL to combine a classification system to perform the Chinese Text chunking task with the small-scale labled Chinese treebank and large-scale unlabled Chinese corpus.

我们通过选取算法理论不同的分类器来定义Co-training方法中的两个"角度"，在选择策略上，我们提出了基于"一致性"的实例添加策略，实验中我们选用基于统计的增益的隐马尔可夫模型和基于转换规则的分类器组合成一个分类体系，并与自我训练方法进行了比较，在小规模汉语树库语料和大规模未带标汉语语料上进行中文组块识别，实验结果要比单纯使用小规模的树库语料有所提高。

To optimize the performance of the hybrid model,it applies k-means algorithm to perform vector clustering in the LSA vector space while the density function is used to initialize the centroid.

研究了潜在语义分析理论及其在连续语音识别中应用的相关技术，在此基础上利用WSJ0文本语料库上构建LSA模型，并将其与3-gram模型进行插值组合，构建了包含语义信息的统计语言模型；同时为了进一步优化混合模型的性能，提出了基于密度函数初始化质心的k-means聚类算法对LSA模型的向量空间进行聚类。

Therefor APSO and DES framework can not only find the better solution but also use fewer iterant times.

所以本研究所提出的演算法，不仅在找到新的最佳解上有好的效能表现，在代数的快速收敛上也有好的效率表现。

Finally, we evaluated the protype system and an open source code anomaly detection system-snort with DARPA 1999, DARPA 2000 data set and the real network flow. The result shows that the algorithms presented in the paper are more improved than snort.

在不同数据集(DARPA1999，DARPA2000)以及实际网络流量上对异常检测原型系统以及开放源码的异常检测系统snort进行对比实验，实验结果表明所提出的算法在误报率、漏报率上都有了很大的提高。

The character information used for recognizing the number includes end-points and trifurcate points, and we can distinguish the different numbers in virtue of them.

数字的大部分信息都集中在骨架线上，所以在细线化图像上提取特征信息，不仅可以大大提高运行速度，而且还能简化算法。

The split between the two groups can hardly be papered over.

这两个团体间的分歧难以掩饰。

This approach not only encourages a greater number of responses, but minimizes the likelihood of stale groupthink.

这种做法不仅鼓励了更多的反应，而且减少跟风的可能性。