重采样

This thesis contributes to geometry compression in the following aspects: Proposes a method of progressive spherical mesh parameterization based on local parameterization by directly projecting the deleted vertices to its neighborhood. Based on spherical mesh paramterization, an adaptive subdivision remeshing method by progressivly redistributing of sampling points is proposed, by this means, the main geometry information of the original mesh can be captured with only a small amount of samples, it can also be used in mesh simplification. By regarding attributes binded on mesh vertices as geometry signals and transferring them onto a uint sphere by means of spherical mesh parameterization, meshes with different attributes are compressed with spherical wavelet successfully with high compression rate. As a kind of lossy geometry compression technique, two mesh simplification algorthimes are also presented in this thesis.

归纳起来，本文的主要贡献有：提出了以直接投影局部参数化方法为基础的网格累进球面参数化方法；在网格球面参数化方法的基础之上，提出了一种基于累进优化布点方案的网格细分采样方法，该方法可以用很少的采样点就可以捕获到网格的主要特征，因此我们也将其用于后面的网格简化工作当中；通过将网格顶点上定义的各种属性看成几何信号，并以网格球面参数化为手段将几何信号迁移到单位球面上，从而成功地实现了用球面小波对各种几何信号的压缩；作为一种有损压缩方法，本文还提出了基于统一距离的网格简化算法和基于重采样的网格简化算法。

The experimental results show the two practical algorithms can preserve inaudibility and they are robust enough to against the different signal processing operations. In addition, the watermarking information can be embedded into the original audio signal adaptively, which also doesnt require the use of the original audio signal for detection. Furthermore, it can resist the different attacks, such as lossy compression (MP3), filtering, resampling and requantizing, etc.

仿真实验结果表明，两种方法都具有较强的鲁棒性和不可感知性，水印嵌入容量的自适应性，提取水印时不需要原始音频信号的参与，且对包括MP3有损压缩、低通滤波、重采样/重量化等多种攻击性试验具有较强的稳健性，是可行的数字音频水印算法。

A novel rendering technology based on surface reconstruction of points is presented,and through local surface analysis,surface parameter extraction and re sampling, fine rendering result at any viewing distance is generated.

提出了一种基于曲面点集重建的渲染技术，能够对输入点集作局部分析，提取曲面的参数，并通过重采样的方法，获得新的、密集的采样点，生成近景的高质量渲染结果。

Inaddition, it gives the linear time re-sampling method to resample the sample set.

最后给出了对样本集进行重采样的线性时间重采样算法。

Based on the proposed algorithm, this paper lays emphasis on the efficient hardware implementation of particle filters on FPGA platform, and presents the hardware architecture of the resample/sample unit and the whole system.

在算法研究的基础上，论文研究了基于FGPA的硬件电路实现方法，给出了系统的整体硬件结构及重采样/采样模块的实现方案，讨论了粒子滤波硬件实现的资源优化及时间优化问题。

The key steps, such as particle initialization, prediction, sequential importance sampling, resampling, are particularly analyzed, and the validity and efficiency of the method are testified by simulation experiment.

着重分析了粒子初始化、预测、序贯重要性采样和重采样等关键步骤，并通过仿真实验分析验证了该方法的正确性和高效率。

Based on the strong relationship about frames near by, the Author brought forward an particle filter algorithm importing weighted sampling about pre-frame in object tracking. The algorithm resolves the trickiness in traditional SIR algorithm which depends on statcmodcl acutely by importing proposal distribution. The algorithm can track object which movement is irregular.

根据相部帧间信息的强关联性，提出一种引入前帧加权采样的粒子滤波目标跟踪算法，解决了传统采样重要性重采样(SIR算法由于引进提议分布而需要严重依赖目标的系统状态模型的问题，可以理想跟踪运动状态不规则的目标。

In this paper w e derived the optimum form of important function by variation method, with that proposed percepts to improve impor tant function; by utilizing the result of delaminateˉ sample theory, the improving direction of reˉsampling was determine d, and with that the double weight selecting partiˉ cle filtering was proposed.

为解决粒子滤波算法在实际应用中出现的滤波发散、粒子退化问题，文中利用变分法证明了重要性函数取法的最优形式，并由此提出重要性函数的改进方案；利用分层抽样原理，确定了重采样方法的改进方向，并由此提出双权重重采样方法。

The Hilbert amplitude envelope of the stator current includes the low frequency of fault characteristic, thus it can reduce the signal length by resample the Hilbert amplitude envelope of the stator current by downsampling, which can avoid spectrum aliasing. Then it can reduce the computation time of the MUSIC Method for the spectrum analysis to the resampled signals for the given bandwidth. Meanwhile combining continuous subdivision Fourier transform and rotor slot harmonics based slip estimation it can in advance estimate the concrete values of the fault characteristic frequency, and then inquiry purposely the fault feature component information in the stator current spectrum in order to accomplish the broken rotor bar fault with high sensitivity and h reliability, which eliminates adverse effect of issuable false peaks to the fault detection and formed a novel scheme to detect broken rotor bar fault.

定子电流Hilbert幅值包络信号包含了低频的故障特征频率调制信号，故通过对定子电流Hilbert幅值包络信号进行降采样率重采样、减少分析信号长度，可避免低频段频谱混叠，再对减少的重采样数据进行具有超分辨率的MUSIC低频段估计，可大大降低MUSIC算法频谱估计时间；运用连续细化傅里叶变换和转子齿槽谐波转差率估计技术，可预知故障特征频率精确值，从而可有目的地查询故障分量，消除MUSIC虚假波峰对故障检测混淆的影响，大大提高了故障检测的灵敏度和可靠性，以此形成异步电动机转子断条故障检测新方法。

Hanrahan is limited to unoccluded space. But occluded environment is quite common in real life scene, so the 5D light field model is proposed in this paper to solve this problem, and the 4D light field is a special case of our model.

我们给出了五维光场模型的定义、表示、采样和重采样过程，并针对五维光场模型采样数据量大的缺点，采用了矢量量化压缩方法，以减少存储空间。

This one mode pays close attention to network credence foundation of the businessman very much.

这一模式非常关注商人的网络信用基础。

Cell morphology of bacterial ghost of Pasteurella multocida was observed by scanning electron microscopy and inactivation ratio was estimated by CFU analysi.

扫描电镜观察多杀性巴氏杆菌细菌幽灵和菌落形成单位评价遗传灭活率。