【手撕算法】图像修复算法整理
通过图像分割和显著性检测这两个系列，我认为在开始写算法之前先放一篇综述整理会更好一些。
综述整理文会包含对该系列大体的介绍，介绍将会分析哪几个经典算法，以及在学习这些算法时我所参考的所有文章资源，我觉的一个算法靠一篇文章绝对无法使大家学会，所以大家可以看看我参考的那些博客，会事半功倍。
github地址：
https://github.com/Yan216/test
图像修复综述
首先辨别一下图像修复，图像复原，图像补全这仨的关系。
图像修复偏重修复划痕之类的这种小破损，而图像补全则偏重修复大块的缺损（包括物体移除之类的），图像复原则是提高图像的成像质量。
本系列的算法主要是图像修复/图像补全这种修复图像的算法。
图像修复(image restoration)包括很多任务，比如降噪/去噪(denoising)，超分辨率重建(super-resolution reconstruction)，图像填充(inpainting)，图像去模糊(deblurring)，JPEG去块(deblocking)等。
本图像修复系列涉及三个图像修复经典算法，分别是：
2003年，Criminisi等人提出的一种基于样本块的修补算法，Criminisi算法通过在待修复区域边缘上选取优先权最高的像素点p，以p为中心构造一个n×n大小的像素块，然后在完好区域寻找与该模板块最相似的样本块，用找到的样本块更新模板块中的待修复信息，最后更新已修复块中像素点的置信度，并开始下一次迭代修复，直至修复完成。
Telea在2004年提出的基于快速行进的修复算法（后面简称FMM算法），它基于快速行进方法。考虑图像中要修复的区域。算法从该区域的边界开始，并进入该区域内部，首先逐渐填充边界中的所有内容。
2009年Barnes提出了PatchMatch 算法。其核心思想是利用图像的连续性（consistence），一个图像patch块A附近的patch块的最近邻最有可能出现在A的最近邻附近，利用这种图像的连续性大量减少搜索范围，通过迭代的方式保证大多数点能尽快收敛。
图像修复博客整理
1
综述类：
图像修复系列-图像修复概述
https://blog.csdn.net/hit1524468/article/details/53444238图像填充修复- 传统算法总结
https://blog.csdn.net/weixin_42386880/article/details/86654947?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522161741227716780271510258%2522%252C%2522scm%2522%253A%252220140713.130102334..%2522%257D&request_id=161741227716780271510258&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2~blog~sobaiduend~default-5-86654947.pc_v2_rank_blog_default&utm_term=%E5%9B%BE%E5%83%8F%E4%BF%AE%E5%A4%8D%E7%AE%97%E6%B3%95&spm=1018.2226.3001.4450图像修复系列-图像修复概述
https://blog.csdn.net/hit1524468/article/details/53444238?utm_medium=distribute.pc_relevant.none-task-blog-2~default~BlogCommendFromMachineLearnPai2~default-9.control&dist_request_id=1328769.23397.16174128011832297&depth_1-utm_source=distribute.pc_relevant.none-task-blog-2~default~BlogCommendFromMachineLearnPai2~default-9.control传统的图像修复方法摘要
https://blog.csdn.net/qq_42641066/article/details/105720473?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522161741363616780261927967%2522%252C%2522scm%2522%253A%252220140713.130102334.pc%255Fall.%2522%257D&request_id=161741363616780261927967&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2~all~first_rank_v2~rank_v29-7-105720473.first_rank_v2_pc_rank_v29&utm_term=%E5%9B%BE%E5%83%8F%E4%BF%AE%E5%A4%8D%E7%AE%97%E6%B3%95&spm=1018.2226.3001.4187图像修复文章汇总
https://blog.csdn.net/yangsn0719/article/details/86643164?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522161741383716780265476150%2522%252C%2522scm%2522%253A%252220140713.130102334.pc%255Fblog.%2522%257D&request_id=161741383716780265476150&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2~blog~first_rank_v2~rank_v29-14-86643164.pc_v2_rank_blog_default&utm_term=%E5%9B%BE%E5%83%8F%E4%BF%AE%E5%A4%8D%E7%AE%97%E6%B3%95&spm=1018.2226.3001.4450图像修复方法总结以及深度学习的论文汇总
https://blog.csdn.net/baidu_33256174/article/details/101173444?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522161741383716780265476150%2522%252C%2522scm%2522%253A%252220140713.130102334.pc%255Fblog.%2522%257D&request_id=161741383716780265476150&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2~blog~first_rank_v2~rank_v29-17-101173444.pc_v2_rank_blog_default&utm_term=%E5%9B%BE%E5%83%8F%E4%BF%AE%E5%A4%8D%E7%AE%97%E6%B3%95&spm=1018.2226.3001.4450图像修复 图像补全_图像修复简介
https://blog.csdn.net/weixin_26752765/article/details/108133477?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522161741383716780265476150%2522%252C%2522scm%2522%253A%252220140713.130102334.pc%255Fblog.%2522%257D&request_id=161741383716780265476150&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2~blog~first_rank_v2~rank_v29-26-108133477.pc_v2_rank_blog_default&utm_term=%E5%9B%BE%E5%83%8F%E4%BF%AE%E5%A4%8D%E7%AE%97%E6%B3%95&spm=1018.2226.3001.4450【论文整理】图像修复(Image Inpainting)必读论文
https://drogozhang.blog.csdn.net/article/details/103329781?utm_medium=distribute.pc_relevant.none-task-blog-2~default~BlogCommendFromMachineLearnPai2~default-2.control&dist_request_id=&depth_1-utm_source=distribute.pc_relevant.none-task-blog-2~default~BlogCommendFromMachineLearnPai2~default-2.control2
Criminisi算法：
图像修复：Object Removal by Exemplar-Based Inpainting 学习笔记
https://blog.csdn.net/weixin_42386880/article/details/85039535?Object Removal by Exemplar-Based Inpainting 翻译
https://blog.csdn.net/u012085234/article/details/86630328复现《Region Filling and Object Removal by Exemplar-Based Image Inpainting》
https://blog.csdn.net/yinaoxiong/article/details/83785122图像修复一：Region Filling and Object Removal by Exemplar-Based Inpainting
https://blog.csdn.net/frank_xu_0818/article/details/18842723 Telea的FMM算法：
opencv图像修复算法cvInpaint(Telea的FMM算法)
https://blog.csdn.net/lsg32/article/details/8932983通过OpenCV库实现基于FMM的图像修复概述
https://blog.csdn.net/weixin_42077074/article/details/110452000?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522161743348416780255279257%2522%252C%2522scm%2522%253A%252220140713.130102334..%2522%257D&request_id=161743348416780255279257&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2~blog~sobaiduend~default-2-110452000.pc_v2_rank_blog_default&utm_term=%E5%9B%BE%E5%83%8F%E4%BF%AE%E5%A4%8DFMM%E7%AE%97%E6%B3%95&spm=1018.2226.3001.4450论文阅读笔记:《PatchMatch: A Randomized Correspondence Algorithm for Structural Image Editing》
https://blog.csdn.net/qq_28223697/article/details/101558502OpenCV中的图像修复算法
https://blog.csdn.net/carson2005/article/details/6844025?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522161743316516780261927879%2522%252C%2522scm%2522%253A%252220140713.130102334..%2522%257D&request_id=161743316516780261927879&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2~all~baidu_landing_v2~default-1-6844025.first_rank_v2_pc_rank_v29&utm_term=An+Image+Inpainting+Technique+Based+on+the+Fast+Marching+Method+&spm=1018.2226.3001.4187快速行进算法（Fast Marching）
https://blog.csdn.net/lusongno1/article/details/88409735?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522161743310416780269864742%2522%252C%2522scm%2522%253A%252220140713.130102334..%2522%257D&request_id=161743310416780269864742&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2~all~sobaiduend~default-1-88409735.first_rank_v2_pc_rank_v29&utm_term=Fast+Marching+Method&spm=1018.2226.3001.4187python 快速行进 算法 图像修补
https://blog.csdn.net/c2a2o2/article/details/79248254?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522161743327616780269879544%2522%252C%2522scm%2522%253A%252220140713.130102334..%2522%257D&request_id=161743327616780269879544&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2~blog~baidu_landing_v2~default-3-79248254.pc_v2_rank_blog_default&utm_term=An+Image+Inpainting+Technique+Based+on+the+Fast+Marching+Method+&spm=1018.2226.3001.44504
PatchMatch算法：
图像修复：PatchMatch与Space-Time completion
https://blog.csdn.net/weixin_42386880/article/details/85177388PatchMatch核心算法（一）
https://blog.csdn.net/CV_YOU/article/details/79083295Patchmatch算法简单实现
https://blog.csdn.net/storMouse/article/details/54667773PatchMatch分析
https://blog.csdn.net/z6491679/article/details/50807689patchmatch
https://blog.csdn.net/haseetxwd/article/details/79718189PatchMatch算法
https://blog.csdn.net/fangjin_kl/article/details/760689815
微信公众号：
图像修复技术
https://mp.weixin.qq.com/s?__biz=MzUzOTkyNzQxOQ==&mid=2247485355&idx=1&sn=6922eb01fde6d81ea6ad1448c9930ddf&chksm=fac1b1bfcdb638a99bfbb7735a2a8ced724b25b79cc8186cc5c84349b79eec143101d65c9201&mpshare=1&scene=23&srcid=0403MmMWhv3NcCls5nDh0Dq1&sharer_sharetime=1617410335963&sharer_shareid=5b6156709f3d332ba42288712d3b8274#rd图像修复简介
https://mp.weixin.qq.com/s?__biz=MzU0NjgzMDIxMQ==&mid=2247497431&idx=2&sn=3ab1ffa4ec974283c3481acaf1bdeca6&chksm=fb55143bcc229d2d5b1f2c6e1f48f7d098eecf1a09893532dcded684ffdb49d68b48df61f405&mpshare=1&scene=23&srcid=0403vhdadv07nH0R43FsASHY&sharer_sharetime=1617410206698&sharer_shareid=5b6156709f3d332ba42288712d3b8274#rd知乎：
修补图像的利器：基于纹理填充技术的Inpainting
https://zhuanlan.zhihu.com/p/77306908纹理合成再谈 - 一种非参数的方法
https://zhuanlan.zhihu.com/p/68430344THE END
今天就到这里啦，明天更新【Criminisi图像修复算法】。