noise

图像检索的一个检索算法基于内容,不错的源码 不错的(A retri algorithm of image retri based on content, good source of good)

本实例用到了第三方组件ICSharpCode.SharpZipLib.dll、itextsharp.dll。

说明：  几种图像融合评价指标，AG,EN,FMI,MI,Qab,Qw,SF,SSIM，即平均梯度AG、信息熵EN、特征互信息FMI、互信息MI、边缘信息保持量Qab、加权融合质量评价因子Qw、空间频率SF和相似性指数SSIM。能较好的描述图像融合结果，稳定性较好。(Several evaluation indexes of image fusion, Ag, en, FMI, MI, qAB, QW, SF, SSIM, namely average gradient AG, information entropy En, feature mutual information FMI, mutual information MI, edge information retention qAB, weighted fusion quality evaluation factor QW, spatial frequency SF and similarity index SSIM. It can describe the result of image fusion well and has good stability.)

说明：  对图像提取同名点，并根据同名点建立delaunay三角网，后续可进行图像配准，或根据三角网进行控制点筛选(The same name points are extracted from the image, and the Delaunay triangulation network is established according to the same name points. The image registration can be carried out later, or the control points can be selected according to the triangulation network)

estimate shape of a noisy image from the shading information

基于小波变换的压缩感知图像处理,及三种重构算法.(Compression based on wavelet transform the perceived image processing and three reconstruction algorithm.)

说明：  基于SURF算法实现图像特征提取与描述，使用RANSAC进行图像细配准(SURF algorithm was used to extract and describe image features, and RANSAC was used for image registration)

载入一副图像并进行平滑处理与高斯或其他核函数进行卷积有效的减少图像信息内容(Loading an image and smoothed with a Gaussian or other nuclear convolution function effectively reduce the information content of the image)

全变分去噪 J=tv(I,iter,dt,ep,lam,I0) 输入: I - 灰度图像, iter - 迭代次数 [1], dt - 时间步长 [0.2], ep - 提升参数epsilon [1], lam - 保真项 lambda [0], I0 - 输入噪声图像 [I0=I] ([]中的是缺省值)( 全变分去噪 J=tv(I,iter,dt,ep,lam,I0) 输入: I - 灰度图像, iter- 迭代次数 [1], dt - 时间步长 [0.2], ep - 提升参数epsilon [1], lam - 保真项 lambda [0], I0 - 输入噪声图像 [I0=I] ([]中的是缺省值))

自己写的matlab版的live-wire算法，效率有点低，不过可用(Written in Matlab version, the live-wire algorithm, a bit low, but available)