ndh

MAD相似性度量准则的改进，实现模板匹配程序(MAD similarity measurement criteria to improve and achieve template matching procedures)

说明：   最简单的差分格式有向前、向后和中心3种。 向前差分：f (n)=f(n+1)-f(n) 向后差分：f (n)=f(n)-f(n-1) 中心差分：f (n)=[f(n+1)-f(n-1)]/2(The easiest difference format forward, backward, and three kinds of centers. Forward differencing: f (n) = f (n+ 1)-f (n) Backward difference: f (n) = f (n)-f (n-1) Central difference: f (n) = [f (n+ 1)-f (n-1)]/2)

原始的criminisi灰度图像修复算法及实验图片(Original Criminisi grayscale image restoration algorithm and experimental images)

说明：  多阈值分割 ，基于otsu生成阈值，有用的可以试试(Multi threshold segmentation, threshold generation based on Otsu, useful can try)

关于超分辨率图像恢复的凸集投影算法，运行结果正确。(code about super-resolution based on POCS)

卡尔曼滤波算法，opencv代码，鼠标跟踪(Kalman filter algorithm, opencv code, mouse tracking)

说明：  文件说明： a1:将二进制文件转为Matalb数据 并批量保存为mat格式 a2:mat数据批量转为图片(RGB三通道） a3:RGB图片批量转灰度图、批量改变图片大小 a4:动图制作 a5:用训练好的网络对图像语义分割(Document description: A1: Convert binary files to Matalb data and save them in batch as mat format A2: mat data batch to picture (RGB three channel) A3:RGB Picture Batch Grayscale, Batch Change Picture Size A4: Motion Mapping A5: Segmentation of image semantics using trained networks)

利用极大似然估计进行盲反卷积复原，通过多帧图像约束，效果较好(Use of maximum likelihood estimates for blind deconvolution restoration by the multi-frame image constraint, the better)

可以直接调用，适用于各种利用contourlet工具箱做为工具的算法(Can be directly invoked, suitable for various algorithms using contourlet toolbox as a tool)

基于双目的深度图获取,根据图像合成深度图左右视点( depth map based on binocular)