Non-local Bayes for Matlab
==========================
Implementation of the non-local Bayes denoising algorithm for matlab.
Compilation
-----------
The main function is `nlbayes.m`. This script runs two steps of nl-bayes denoising.
You need to modify the code of the script to change parameters, input image, etc.
Notes
-----
For simplicity, this version has some differences with respect to the [IPOL
implementation][1]:
* We work in both steps with RGB patches. The IPOL implementation does a different
handling of color in the first step.
* We consider always `prms.np` similar patches. The IPOL implementation can use more
similar patches in the second step, if the distances are smaller than a threshold.
* The user needs to provide all parameters.
* We add a parameter `prms.r` to reduce the rank of the _a priori_ covariance matrix.
Only the `r` leading eigenvectors of the covariance matrix are kept.
It is disabled by default.
[1]: http://www.ipol.im/pub/art/2013/16/
Visualization of Patch Groups Computed by NL-Bayes
==================================================
The function `patch_group_image.m` is an interactive tool for visualizing the
patch groups built by the (video) nlbayes algorithm.
Parameters
----------
The following parameters can be specified by modifying the code:
* patches size
* search region
* number of similar patches
Noise values
------------
The noise value can be controlled using the interactive tool.
Names of the image files
------------------------
The function dumps images with the result with the naming convention:
`patch_group_[IMAGE]_[REFx]_[REFy]_s[SIGMA]_[DISTA]_[WSZ]_[PSZ]_[NP]_[CODE].png`
where
* `IMAGE` : name of the image where we took the patches from
* `REFx ` : x coordinate of top-left pixel in reference patch
* `REFy ` : y coordinate of top-left pixel in reference patch
* `SIGMA` : noise std. deviation
* `DISTA` : patch distance used (e.g. l1 or l2)
* `WSZ` : (spatial) size of search region
* `PSZ` : (spatial) patch size
* `NP` : number of similar patches
* `CODE ` : a 4-letter code indicating what the file shows (see bellow)
For example:
`patch_group_traffic_coor_545_290_s20.png`
shows the coordinates (coor) of the similar patches for a reference patch at
position (545,290) in the image `traffic` with sigma = 20.
What is shown in each file?
---------------------------
For each reference patch and each noise value we show several images
identified with a 4 letter abbreviation:
* `coor`: coordinates of the top-left corner of the nearest neighbors shown in
the search region. We use the following color code:
- RED : nearest neighbors from 1 to 5
- GREEN: nearest neighbors from 6 to 45
- BLUE : rest of the nearest neighbors
* `nisy`: set of noisy similar patches found in the search region.
The reference patch is in the top-left corner. The nearest neighbors
are ordered from left-to-right, top-to-bottom.
* `orig`: the noiseless version of the patches in 'nisy' image. Note that the
similar patches were found according to the distance between the
noisy patches. The noiseless version is shown to see how the set
of similar patches degrades when the noise increases.
* `pcas`: mean patch and eigenvectors of the sample covariance matrix. The
patch in the top-left corner is the mean patch. The eigenvectors
are shown ordered by decreasing variance from left-to-right,
top-to-bottom. To visualize the eigenvalues, we add a color-coded
border around each eigenvector.
- A GREEN border, indicates an eigenvalues larger than sigma^2. The
intensity of the green is proportional to `\sqrt{\lambda_i} - \sigma.`
- A RED border shows an eigenvalue smaller than sigma^2 with intensity
proportional to `\sigma - \sqrt{\lambda_i}.`
* `eigs`: plot of the square root of the eigenvalues, compared to sigma. In blue
we plot the positive eigenvals, and in red the negative.
Don't pay attention to the legend in the plot, it's wrong!
没有合适的资源?快使用搜索试试~ 我知道了~
机器视觉算法库-去噪算法.7z
共515个文件
m:263个
bmp:69个
png:62个
1.该资源内容由用户上传,如若侵权请联系客服进行举报
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
版权申诉
0 下载量 56 浏览量
2023-07-04
12:01:14
上传
评论
收藏 67.82MB 7Z 举报
温馨提示
机器视觉算法库-去噪算法.7z
资源推荐
资源详情
资源评论
收起资源包目录
机器视觉算法库-去噪算法.7z (515个子文件)
+generate 1B
goldhill512.bmp 257KB
peppers512.bmp 257KB
baboon512.bmp 257KB
lena512.bmp 257KB
barbara512.bmp 257KB
goldhill512.bmp 257KB
peppers512.bmp 257KB
baboon512.bmp 257KB
lena512.bmp 257KB
barbara512.bmp 257KB
zebra256.bmp 87KB
zebra256.bmp 87KB
lena256.bmp 65KB
peppers256.bmp 65KB
manmade256.bmp 65KB
bridge256.bmp 65KB
cameraman256.bmp 65KB
dog256.bmp 65KB
lena256.bmp 65KB
peppers256.bmp 65KB
manmade256.bmp 65KB
bridge256.bmp 65KB
cameraman256.bmp 65KB
dog256.bmp 65KB
couple256.bmp 65KB
0.bmp 65KB
6_NASNLM.bmp 65KB
2_ACWMF.bmp 65KB
4_NAFSM.bmp 65KB
1.bmp 65KB
3_DBA.bmp 65KB
5_NASEPF.bmp 65KB
1_MF.bmp 65KB
barbara256.bmp 65KB
result3.bmp 65KB
result5.bmp 65KB
couple256.bmp 65KB
edge0.bmp 65KB
result1.bmp 65KB
edge1.bmp 65KB
result2.bmp 65KB
barbara256.bmp 65KB
result6.bmp 65KB
result4.bmp 65KB
zebra128.bmp 23KB
zebra128.bmp 23KB
peppers128.bmp 17KB
barbara128.bmp 17KB
bridge128.bmp 17KB
manmade128.bmp 17KB
lena128.bmp 17KB
peppers128.bmp 17KB
barbara128.bmp 17KB
bridge128.bmp 17KB
manmade128.bmp 17KB
lena128.bmp 17KB
edge2.bmp 8KB
zebra64.bmp 7KB
zebra64.bmp 7KB
barbara64.bmp 5KB
bridge64.bmp 5KB
manmade64.bmp 5KB
peppers64.bmp 5KB
lena64.bmp 5KB
barbara64.bmp 5KB
bridge64.bmp 5KB
manmade64.bmp 5KB
peppers64.bmp 5KB
lena64.bmp 5KB
bm3d_wiener_video_c.dll 272KB
bm3d_thr_video_c.dll 236KB
bm3d_thr_colored_noise.dll 212KB
bm3d_thr.dll 196KB
bm3d_thr_video.dll 192KB
bm3d_thr_sharpen_var.dll 184KB
bm3d_wiener_video.dll 180KB
bm3d_wiener_color.dll 156KB
bm3d_wiener_colored_noise.dll 152KB
bm3d_thr_color.dll 148KB
bm3d_wiener.dll 128KB
.DS_Store 6KB
.DS_Store 6KB
.DS_Store 6KB
NL Means sigma 20 h 0.15.fig 920KB
NL Means sigma 20 h 0.25.fig 920KB
NL Means sigma 20 h 0.55 with integral.fig 920KB
NL Means sigma 20 h 1.fig 919KB
NL Means sigma 20 h 15.fig 875KB
NL Means sigma 5 h 0.25.fig 769KB
NL Means sigma 5 h 0.55.fig 768KB
.gitattributes 378B
.gitattributes 378B
.gitignore 649B
.gitignore 574B
.gitignore 19B
ill_DMS_Graph.jpg 660KB
noise_example.jpg 149KB
LICENSE 1KB
CBM3D.m 27KB
共 515 条
- 1
- 2
- 3
- 4
- 5
- 6
资源评论
应用市场
- 粉丝: 461
- 资源: 3815
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功