power-spectrum

此程序代码实现超分辨率图像的重建、复原 联合开发网 - pudn.com

区域生长法获取图像边界 必须有种子 可以有很多方法 这里直接鼠标点击(Region growing method to obtain the image the border must have a lot of methods of seed can have a direct mouse click here)

计算两个离散变量x和y的联合熵z = H（x，y）。 输入：x，y：两个相同长度的整数向量输出：z：联合熵 可以使用，有利于学习目的(Compute joint entropy z=H(x,y) of two discrete variables x and y. Input: of x, y: two integer vector of the same length Output: z: joint entropy good for learning purposes)

kinect扫描到的人体分离出来，并以二值图显示，且可以保存在本地，丰富的注释帮助理解(kinect body scan to be separated, and binary image display and can be saved locally, rich annotation to help understand)

POCS算法 基本算法 肯定能用经过多次试验保证质量希望能够帮助大家 大家加油(POCS algorithm is sure to be able to use a number of tests to ensure that the quality of the hope can help everyone to refuel)

论文介绍了一种SFS 算法的参数估计及其实现。它在考虑自遮掩影响的情况下, 有效地估计了SFS 算法中涉及的各种控制参数, 并引入亮度约束、灰度梯度约束和可积性约束, 计算出表面高度和表面向量, 实现三维重构。最后还指出了在Matlab 中实现需要注意的问题。(The paper introduced a SFS algorithm parameter estimation and its implementation. It under the condition of considering the covering effect, estimates the SFS algorithm effectively involved in all kinds of control parameters, and introduces brightness constraint, gray level gradient constraint and the integrability of constraints, calculate the surface height and surface vector, realize the three-dimensional reconstruction. Finally, points out the problems requiring attention is realized in Matlab. )

水平集分割图像的一种算法，属于比较前沿的一种图像分割方法。(A level set image segmentation algorithm, a relatively cutting edge of a segmentation method.)

对比五种模糊图像复原算法，包括：维纳滤波算法、约束最小二乘滤波算法、Richardson-Lucy（RL）算法、循环边界算法和最优窗算法。(introduces the five kinds of image restoration algorithms, including wiener filtration, constrained least squares filtering, Richardson-Lucy, cyclic boundary algorithm and optimal window algorithm.)

程序设计步骤： 1、去均值 2、计算协方差矩阵及其特征值和特征向量 3、计算协方差矩阵的特征值大于阈值的个数 4、降序排列特征值 5、去掉较小的特征值 6、去掉较大的特征值(一般没有这一步） 7、合并选择的特征值 8、选择相应的特征值和特征向量 9、计算白化矩阵 10、提取主分量(Programming steps: 1 to mean 2, calculation of covariance matrix and its eigenvalues and eigenvectors 3, the calculation of covariance matrix eigenvalues greater than the threshold number of 4, in descending order of eigenvalue 5, remove the smaller eigenvalue 6, remove the larger eigenvalue (usually not this step) 7, the combined choice of eigenvalues 8, select the appropriate eigenvalues and eigenvectors 9, computing whitening matrix 10, principal component extraction)

说明：  基于VC++的图像的DFT变换和DCT变换，并可实现两者间的对比转换(VC++ based on the image of the DFT transform and DCT transform, and to achieve the contrast between the two conversion)