变换点

And the matching point pairs are obtained by comparing the Euclidean distance of the Zernike moments,which are computed around a circular neighborhood centered at the feature point.Then the transform parameters are computed through the correct mapping model.Finally,the weighed average method is used to achieve image mosaic.

该算法利用曲率尺度空间角点检测法提取特征点，通过比较各特征点圆形邻域Zernike矩的欧式距离得到匹配点对，然后利用正确的映射模型计算出变换参数，最后采用加权平均法得到图像的拼接结果。

The method for the point sets selection of point Hough transform and the scan direction for image of searching the point sets were improved by using the geometrical characteristic of the concentric circles:The center of the concentric circles was obtained in the accumulative array firstly,radius was obtained by the mean of distance between the circle center and every edge points afterwards,both the circle center and the edge of the concentric circles were detected accordingly.

利用同心圆的几何特点对点Hough变换边缘点的点组选取方法和搜索点组的图像扫描方向进行了改进：首先由累加阵列确定同心圆的圆心，然后由圆心到各个有效边缘点距离的均值求出半径，从而检测出同心圆的圆心与边缘。

Each cell was mapped by the center point on the cell of transverse section with extended up and down for four pixel(nine-pixel lengths in total) and two pixels in width,and thereafter line convoluted dot map was obtained.

把每个纤维细胞都用其中心的一个长9个像素点、宽2个像素点的线段点图表示，进行傅立叶变换，其能量光谱图在水平和垂直方向上的频率分布，除零值以外的最大值，即为细胞径向、弦向直径。

Among many implement schemes of decoding, the one using chips is the fastest one. Specific decoding chips like FPGA or ASIC have parallel processing structures, which make them be much fit for being high performance data processing processors like Digital Filter, Fast Fourier Transform, Discrete Cosine Transform, etc, and also be the key part of high performance data processing systems. The implementation is synthesizable Verilog code at the RTL level. It is suitable for real time video applications.

在MP3 解码电路中，模块IMDCT 由三个子模块组成：IDCT子模块、加窗函数子模块、输出数据的叠加子模块。N/2 点的输入Xk 经IDCT 变换得到N 点的输出xi 的公式如下：将输出根据不同的块类型(有四种块类型，分别为普通、开始、结束、短型)采用不同的窗函数加窗，得到一个36 个点的输出(对于块类型为短型的块，单个小块的输出是12 个点，每三个小块组合成一个36 个点的输出)。

The main contributions of the thesis are as follows: Using invariants and matrix theory, especially the special properties of eigenvalues and eigenvectors of the so called irreducible matrix, we put forward several effective new algorithms for the complete matching problem under Euclidean transformation, similarity transformation and general affine transformation.

论文的主要贡献如下：利用不变量和矩阵理论，特别是巧妙地利用了不可约矩阵的特征值和特征向量的特殊性质，提出了求解欧氏变换、相似变换和一般仿射变换下完全匹配情形的点模式匹配问题的几种非常有效的新算法。

The main work and innovation1 In general , a image compression method has different effect to different image . Some are fit for image which has much high frequency . Some are fit for image which has much low frequency .This paper analysis the applicability of fractal algorithm . In this dissertation , an improved algorithum is given . The basic algorithum without eight change and the joint use of quantitative coding , The compression ratio and PSNR can be adjusted , and it can get more quickly speed of coding .2 To a group of exact fractal code , the decoding image can be any resolution .With this characteristic ,choose the proper domain block and range block,this paper get an algorithum which can get higher decoding PSNR and speed than the basic fractal algorithum.3 Give a new method to calculate the scale factor . The calculation of basic algorithum is complex and enormous . The factor of scale is an important variable of the basic algorithm. Now there are two methods to calculate the factor of scale .A new method was given to calculate the factor of scale. The calculation load and complexity are reduced by the new method. Compared with the basic algorithm , the method can reduce coding time.4 The basic fractal algorithum is based on the Collage theorem and contraction mapping theorem . A mathematic model based on basic fractal algorithum is given .

本文的主要工作及创新点主要包括以下几点：1一般情况下，一种图像压缩的方法对不同特点的图像，压缩效果会有所不同，本文利用分形压缩编码对不同特点的图像进行编码，分析了分形编码对不同图像的编码适用性，利用去掉8种变换的分形编码算法与简单的均匀量化编码方案相结合，得到了一种可以调节压缩比与解码质量的分形编码方案，该提高编码速度6倍以上。2分形编码的特点之一是分形解码的分辨率无关性，即对于一组分形编码可以将其解码到任意分辨率下，本文利用分形编码的这一特点，结合不同的量化编码方案，选择合适的定义域块与值域块及适当的编解码分辨率，得到了一种提高解码质量，编码速度提高12倍以上的加速算法。3比例因子是基本分形算法中的一个重要变量，目前有两种计算比例因子的方法，文中给出了一种新的求比例因子的方法，该方法简化了求比例因子的计算，从整体上减小了基本分形算法的计算量，简化了算法4分形压缩的数学基础是不动点定理与拼贴定理，本文根据基本的自动分形编码方法，建立数学模型对其进行推导，得到了用普通数学的极限理论解释分形编码的数学表达式，从而对由基本分形编码得到的解码图像中的像素点的值给出了确定的表达式。

First, the proposed scheme uses contourlet transform to extract multidirectional and multiscale texture information, and then obtains the feature points invariant to the affine transformation from the low and middle directional subbands by Harris-Affine detector. Second, the feature regions are adaptively computed by the feature scale of the local structure and normalized by U transform. The watermarking is embedded adaptively in those characteristic regions which are normalized. Finally, by vector quantization, several copies of the watermark are embedded into the no overlapped local feature regions in different directional subbands.

首先，用Contourlet变换提取出多尺度、多方向的纹理信息；再用Harris-Affine检测算子从变换域的中、低频方向子带中提取出仿射不变特征点，结合自适应局部结构的特征尺度确定特征区域，并用U变换对其归一化处理，水印就自适应地嵌入到归一化后的区域中；水印嵌入采纳矢量量化的策略，将水印信息重复嵌入到不同方向子带、多个不相交的局部仿射不变特征区域。

The synthesis scheme we developed can deal with changes in visibility, produce perspectively correct views and handle the distortion around the circumferences in the synthesized views. In our scheme, the area without corresponding points was first determined by using Fourier transformation in order to deal with distortion problem, within the mutual overlapped area the z-buffer values of matched key points were computed so that the visibility in novel views can be handle correctly, lastly the common points in new views are rendered through epipolar geometry constraint.

算法的步骤为：利用傅立叶变换方法确定相邻源图像无对应点的区域；然后，利用尺度空间和相位一致原理在共同区域提取及匹配边缘上的关键点，计算其深度值，并在新图像中由该值判别虚拟图像中的可见性；再由关键点恢复极线约束，通过极线间的位置和颜色插值生成虚拟图像中有对应区域的一般点；在确定极线位于无对应点区域段后，利用加权平均或放大插值的方法实现了源图像无对应点区域在虚拟图像中的自然过渡。

This paper extends the solution methods for 2-D projective transformation matrix from 4 points corresponding to variety combinations of points and lines corresponding, such as 4 points, 4 lines, 1 point and 3 lines, 3 lines and 1 point, and 2 points and 2 lines, etc.

该文把求解2-D射影变换矩阵的方法从单纯的四点对应扩展到多种组合形式的四元素对应当中，如：4点、4线、3点1线、3线1点和2点2线等。

The feature points from two images were extracted by Harris corner detector, then the corresponding feature point pairs were got by the cross-correlation of the gray scale around the feature points, and the mismatched pairs were eliminated based on the Eulerian distance of the transformation parameter vector.

提出了一种基于特征点的图像拼接算法，首先利用Harris角点检测提取两幅图像中的特征点，用特征点邻域灰度相关性将特征点进行匹配，并在此基础上根据变换参数向量的欧式距离对误匹配进行剔除，进而实现图像拼接。

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。