Firstly, wavelet transform was made en sample images through quadric B-spline wavelet filters, and multi-resolution edge was extracted. Then moment invariant of edge image was extracted, which was taken as the characteristic vector. Finally, the wavelet neural network was used to classify and recognize the target.
首先采用二次B样条小波滤波器组对样本图像进行小波变换，提取多尺度边缘，然后提取不变矩，以此作为图像的特征向量，最后应用小波神经网络进行分类识别。
If the space shift invariance is replaced by the time shift invariance, the high resolution frequency estimation can be obtained using a single vector sensor.
如果将空间的位移变成时间的延迟，单个矢量传感器可以实现高分辨率的频率估计。
Overall aims and objectives of our research project are: i Construct retrovirus vector and producer cell line belonging to our own laboratory and acquire the basic information. ii Compare the therapeutic consequences between intracranial tumor and subcutaneous tumor in rat abdomen, and between injection of producer cells inside abdomen tumor in one point and in multi-points pattern. iii Monitor the growth or regression of tumors with high resolution and contrast enhanced MRI for investigating whole process of gene therapy in vivo. iv Propose the manchanism of bystander effect according to our experimental finding.
我们的研究计划主要侧重于：建立自己实验室的逆转录病毒载体及包装细胞系，获得基础研究的第一手资料；对比颅内肿瘤与颅外肿瘤、包装细胞单点注射与多点注射的基因治疗方式的疗效差别；应用高分辩率的强化与非强化MRI全程监测整个体内基因治疗的试验过程；探讨旁效应的机理，提出实验依据。
Firstly, SPNs are extracted from chest CT images using the interactive segmentation, and the multi-resolution histograms of SPNs are directly calculated to receive a high-dimensional feature sample set with spatial information of SPNs. Then, the classifier for differentiating benign and malignant SPN is constructed by using a Support Vector Mechine. Finally, the performance of classification is evaluated by testing the trained SVM with the test sample set.
采取交互式分割方法从胸部CT图像中提取出SPN；直接计算SPN图像的多分辨率直方图得到768维空间信息特征样本集；然后，充分利用具有处理高维数据集优势的支持向量机构造SPN良恶性分类器；最后，通过测试样本集对经训练后的SVM分类器进行测试以评价分类性能。
Firstly, SPNs are extracted from chest CT images using the interactive segmentation, and the multi-resolution histograms of SPNs are directly calculated to receive a high-dimensional feature sample set with spatial information of SPNs. Then, the classifier for differentiating benign and malignant SPN is constructed by using a Support Vector Mechine.
采取交互式分割方法从胸部CT图像中提取出SPN；直接计算SPN图像的多分辨率直方图得到768维空间信息特征样本集；然后，充分利用具有处理高维数据集优势的支持向量机构造SPN良恶性分类器；最后，通过测试样本集对经训练后的SVM分类器进行测试以评价分类性能。
Even if flying outside the atmosphere, in order to ensure hit-kill the target, the common radar seeker had better not be used either. It is possible for the TBM defense interceptor to hit to kill the target or to kill the target to draw support from a strengthen kill-device with a radar seeker of high range resolution ratio or a high precision LWIR stare-image seeker, by gas thrust vector control system or gas thrust vector control/aerodynamic control system, to make use of optimum guidance law. The statically stable interceptor with gas dynamic control system in the atmosphere has the advantages that the fuel consumption of the attitude control engine is small and the error of body's angular velocity is little in the period of stabilization control. To achieve long distance ascent-phase intercept, such as the ground-launched TBM defense interceptor located in the point TBM attacking, very high demands are raised to the TBM defense system and the TBM defense interceptor, because the increase in interceptor mass is about one percent if the stand off range of the interceptor increases one kilometer, and the increase is about six percent if the intercept altitude decreases one kilometer or the reaction time of the TBM defense system increases one second for flight time of 60 to 90s, altitude of 50 to 80km and stand off range of 300 to 600km.
理论研究与数字仿真结果表明：大气层内采用普通雷达导引头的反TBM拦截器无论采用空气动力控制还是直接侧向力控制，都难以实现与目标的直接碰撞，即使在大气层外飞行，要确保与目标直接碰撞，也不宜采用普通雷达导引头，而采用高距离分辨率雷达导引头或高精度LWIR凝视成像导引头，利用直接侧向力控制或直接侧向力／空气动力复合控制方式，运用最佳制导规律的反TBM拦截器实现与目标的直接碰撞或利用杀伤增强装置杀伤目标是可能的；大气层内燃气动力控制的静稳定拦截器稳定控制时具有姿控发动机能耗少、弹体角速率误差小的优点；远距离拦截处于上升段TBM时，如采用在TBM攻击区域的地基拦截方案，对TBM防御系统及反TBM拦截器均提出了很高的要求，因为射程每增加一公里，拦截器的质量将增加百分之一左右，拦截高度每降低一公里或TBM防御系统反应时间每增加一秒，反TBM拦截器的质量将增加百分之六左右（针对射程300～600km，拦截高度50～80km，飞行时间60～90s的情况）。
Using the method, the image blocks in different resolution levels can be well represented by the centroids of the vector subsets.
用这种方法，使得在不同分辨率和不同频段的子图像块能用其矢量集的质心来表示以达到压缩的目的。
Firstly, the eigen vector matrix and eigen value of Jacobian matrix of governing equations for the real gas flows in general curvilinear coordinate system are deduced. And a numerical method with high resolution TVD schemes for solving the governing equations of real gas flows is developed.
首先对真实气体流动的数值模拟方法进行了研究，推导了任意曲线坐标系下真实气体流动控制方程雅克比系数矩阵的特征向量矩阵和特征值；以此为基础，构造了求解真实气体流动控制方程的高分辨率TVD数值格式。
Clear images of particles were obtained by combining 10x microscope, 14-digit grey scale PCO1600 camera, 3μm fluorescence tracer particles and light filtering lens allowing 610nm red light permeation. The flow field velocity vector patterns and streamlined diagrams of high resolution were obtained by using image processing techniques.
试验使用10倍显微物镜、14位灰阶PCO.1600相机、3 mm荧光示踪粒子和仅允许610 nm红光透过的滤光镜相配合获取清晰的粒子图像，通过图像处理技术得到高分辨率的流场速度矢量分布图和流线图。
In the proposed method, a classic Douglas-Peucker algorithm and a new vector height classification algorithm are adopted to retrieve spatial data with diverse resolution and then the new level search tree is introduced as the incremental data buffer storage model. Finally, vector maps are created using SVG technique.
该方法运用DP算法和失高分级算法来提取不同分辨率的数据，以等级搜索树作为数据缓冲增量存储模型，采用SVG技术生成矢量地图。
Every Architect, Designer, Engineer and Contractor using Photoshop ought to know these essential concepts: how images are stored, how measured drawing scale relates to image resolution, learn a host of digital photo retouching techniques, and get the most out of powerful vector tools.
每一个建筑师，设计师，工程师和承包商利用Photoshop应该知道的这些基本概念：怎样影像储存，如何测量制图规模涉及到的图像分辨率，学习数码相片修饰技巧，并获得最大的强大载体工具。
In disturbance detection module, the detection of various disturbances is implemented by adaptive linear neuron and the detected disturbance signals are sent into classification module; by use of discrete wavelet multi-resolution analysis and Parseval`s theorem the energy distribution features under different scales are extracted, meanwhile by use of fractal geometry the local variance dimensions are extracted; combining these two items, the characteristic vector of disturbance signal is composed.
在扰动检测模块中，采用自适应线性神经元实现了对各种扰动的检测，将检测出的扰动信号送入分类模块，采用离散小波多分辨率分析提取不同尺度下的能量分布特征，同时采用分形几何学提取局部方差维数，将二者结合共同构成扰动信号的特征矢量。
The benefit of changing graphics from raster to vector is you do not need to implement anything special for different resolution.
改变的光栅矢量图形的好处是你不需要执行任何特殊的不同的分辨率。
After studying the theory of combined signal processing with acoustic pressure and particle velocity, a method of direction estimation underwater, which is based on intensity measurement using an acoustic vector sensor, is presented. In the low frequency field with kinds of line-spectra, azimuth angle estimation is achieved based on the probability density statistics and bar graph approach. It can preferably resolve the problem of left/right bearing resolution and estimate the direction of space.
为了对水下目标进行定位，基于声压与振速联合信号处理的理论，研究了复声强法对水下目标信号的定向技术，并使用概率统计和直方图的理论对目标的方位角和俯仰角值进行了估计，在线谱分量存在的低频段对估计角度进行了概率统计，较好地解决了左右舷的分辨问题，实现目标的全空间无模糊定向。
Discussed results show that the feature vector is invariable for images in complex transformations on translation, rotation, noise adding and differences in resolution, and can be used as the features in the application of imaging matching location.
综合得出，在一定的条件下，该特征向量对平移、旋转、加噪、分辨率变化等多种复合变换下的图像具有不变性，可以作为成像匹配定位应用的特征量。
In this paper, the author adopted different methods to discuss resolution including geometry optics, Gaussian beam and vector diffraction integration. We used ray tracing to analyze focal area`s position and made further analysis of the affection with horn position`s change. Through analyzing and experimenting, we discussed beam feeding network and multiple beam`s coupling. We analyzed the affection on resolution induced by horns intervals`change, furthermore, we proved this affection by experiment. At last, we discussed the spillover and illumination level. With experiments of single beam and multiple beam lens antenna, we got results that their specifications achieved the project goals.
本文采用几何光学法、高斯波束和矢量衍射积分法分析了焦面上的分辨率；用射线追迹对馈源放在不同位置时，对焦斑位置的影响进行了分析；多个馈源之间的位置变化对分辨率的影响进行了仿真，并在试验中得到了满意的结果；分析了多波束天线的不同阵列对分辨率的影响；讨论了馈源尺寸变化对分辨率的影响；对多波束天线的馈电网络、馈源间的耦合关系进行分析和实验验证；分析了透镜天线的溢出效率和边缘电平。
Feature Vector Based Chinese Coreference Resolution Research and System Implementation
基于特征向量的中文指代消解研究与系统实现
The measurements of the radio spikes with high time and frequency resolution by the spectrometer in Beijing Observatory may provide a self consistent diagnosis for the ambient plasma parameters (magnetic field, density, temperature, wave vector, and the properties of nonthermal electrons).
利用北京天文台高时间和高频率分辨率的射电频谱仪对射电尖峰的测量，可以对背景等离子体参数进行的自洽诊断（磁场，密度，温度，波矢，及非热电子的性质）。
The conventional minimum variance beamformer (MVB) is known to have better resolution and much better anti-interference capability when the array steering vector is accurately known.
在阵列导向矢量不存在误差的条件下，传统的最小方差波束形成器（MVB）具有良好的分辨精度和抗干扰能力。
The results indicate that ESPRIT algorithm based on vector sensor can obtain higher resolution frequency estimation than that based on pressure hydrophone.
结果表明基于矢量传感器的算法比基于声压传感器的算法具有更高的频率估计精确度。