sfm_chanvese_demo

斑点噪声的去除方法，主要适用sar图像，比较适合开题使用(Speckle noise removal methods, mainly for sar image, more suitable for the use of open questions)

semi-supervised learning 半监督局部线性嵌入(semi-supervised learning )

说明：  比SSIM更新的图像质量评判算法VSNR(image quality assessment VSNR and its matlab code)

并行磁共振成算法GRAPPA，希望对大家有帮助，内有K空间数据。(Generalized Autocalibrating Partially Parallel Acquisitions (GRAPPA))

对立体视觉的双目图像进行极线校正，以实现双目致密匹配(Of three-dimensional visual image of the binocular epipolar rectification to realize dense binocular matching)

图像分割是把图像划分成具有实际意义的互补交迭的区域的集合。在图像分割之前，图像区域的数目未知，而在分割后各个区域同时满足均匀性和连通性的条件，故图像分割是一个复杂的过程，目前大多数研究都是针对某一类型图像或者某一具体应用的分割。图像阈值化分割的基本思想是确定一个阈值，然后把每个像素点的像素值和阈值相比较，根据比较的结果把像素划分为两类，前景(1)或背景(0)。该方法的关键是确定一个最优的阈值。常用的阈值确定方法有直接门限法、类间最大方差法(otsu法)、分水岭算法、最小误差法、最大熵法等。该段代码就是图像的阈值分割。(The image segmentation is divided into the practical significance of the emerging area of complementary set. In the image segmentation, before the number of image region, the unknown, and after each area in division at the same time satisfy the uniformity and connectivity condition, so the image segmentation is a complicated process now, most research are for a certain type image or a particular application of the division. Image threshold segmentation of the basic idea is to determine a threshold, then put each pixel threshold value of the pixels and compared, according to the result of the comparison the pixel divided into two categories, the prospect (1) or background (0). This method is the key to determine an optimal threshold. Commonly used method to determine the threshold value of direct threshold method, the largest of the variance between method (otsu method), watershed algorithm, the minimum error method, the maximum entropy method, etc. When the code is image threshold segm)

先将图像分成一个个小区域后在进行合并，最后得到要提取的目标(First image into a small area after carrying out the merger, the final goal has been to extract)

基于稀疏分解的欠定盲源分离，要求信号具有一定的稀疏性。(Based on sparse decomposition due to blind source separation, the signal has a certain sparseness.)

说明：  利用匹配滤波器实现图形识别，包括程序和说明文档(Graphic Recognition Using Matched Filter)

这写程序是对加水印之后图像的攻击，包括加噪声，各种滤波，剪切，压缩攻击和旋转等。(This writing process is a plus watermark image after the attacks, including the increase in noise, various filtering, shear, compression and rotation attacks.)