数学建模运动模糊图像的复原

奇异值分解降噪程序，压缩包里有程序的详细说明，内附有word文档的说明(SVD noise reduction program, compression bag has a detailed description of the program, enclosing a word document description)

说明：  可以将txt文件中的某一列数据进行提取出来，然后进行图像制图(You can extract a column of data from the txt file and then do image mapping)

【实例简介】

2012 CVPR 文章：Segmentation Using Superpixels A Bipartite Graph Partitioning Approach对应的代码(2012 CVPR paper：Segmentation Using Superpixels A Bipartite Graph Partitioning Approach)

说明：  有四个M文件，可以用于图像处理中的图像细化，有较好的效果(There are four M documents, can be used for image processing of image thinning, there are good results)

VC++Digital image processing

水平集(level set)的基本思想是将界面看成高一维空间中某一函数ψ（称为水平集函数）的零水平集，同时界面的演化也扩充到高一维的空间中。我们将水平集函数按照它所满足的发展方程进行演化或迭代，由于水平集函数不断进行演化，所以对应的零水平集也在不断变化，当水平集演化趋于平稳时，演化停止，得到界面形状。(The level set (level set) the basic idea is to interface as a higher dimensional space in a function (called a level set function) of the zero level set, and the interface evolution also extended to higher dimensional space. We evolve or iterate the level set function according to the evolution equation it satisfies. Because the level set function keeps evolving, so the corresponding zero level set is also changing. When the level set evolves to a stable state, the evolution stops, and the interface shape is obtained.)

bandelet 去噪，包含blsgsm和平移不变去噪，不过对大图像需要的内存比较大(bandelet denoising, including blsgsm and translation invariant denoising, but for large images need more memory)

一篇关于总结了行为识别各种资源的论文，涵盖了通常的数据集、兴趣点检测，特征提取的方法以及识别检测的算法(A paper that summarizes the various resources of behavior recognition, covering the normal data set, the interest point detection, feature extraction methods and the identification detection algorithm)

说明：  代码论文Active Contours Without Edges(code paper Active Contours Without Edges)