Linux/Unix>> Manuscripts-restoration>> 返回
项目作者： simoneparisotto

项目描述 ：
This is the companion software for the paper "Unveiling the invisible - mathematical methods for restoring and interpreting illuminated manuscripts".
高级语言： C
项目地址: git://github.com/simoneparisotto/Manuscripts-restoration.git
创建时间： 2018-04-28T22:12:48Z
项目社区：https://github.com/simoneparisotto/Manuscripts-restoration
开源协议：BSD 3-Clause "New" or "Revised" License
下载


Digital restoration of Illuminated Manuscripts

Authors of this software: Simone Parisotto and Luca Calatroni

Other authors and collaborators: Carola-Bibiane Schönlieb, Stella Panayotova, Paola Ricciardi

Version 1.0

Date: 22/02/2018

This is a companion software for the journal article:

  1. L. Calatroni, M. D’Autume, R. Hocking, S. Panayotova, S. Parisotto, P. Ricciardi, C.-B. Schönlieb, 
  2. "Unveiling the invisible - mathematical methods for restoring and interpreting illuminated manuscripts"
  3. Heritage Science, Springer International Publishing (2018) 6: 56. DOI: 10.1186/s40494-018-0216-z

Please use the following entry to cite the article:

  1. @Article{CalAutHocPanParRicSch2018,
  2. author="Calatroni, Luca and d'Autume, Marie and Hocking, Rob 
  3. and Panayotova, Stella and Parisotto, Simone 
  4. and Ricciardi, Paola and Sch{\"o}nlieb, Carola-Bibiane",
  5. title="Unveiling the invisible: mathematical methods for restoring and interpreting illuminated manuscripts",
  6. journal="Heritage Science",
  7. year="2018",
  8. volume="6",
  9. number="1",
  10. pages="56",
  11. publisher="Springer International Publishing",
  12. doi="10.1186/s40494-018-0216-z",
  13. url="https://doi.org/10.1186/s40494-018-0216-z"
  14. }

Code description: segmentation and Inpainting step

The workflow consists in two steps: the recognition of damaged areas in manuscripts and the inpainting restoration.

  • 1st step: run Matlab file for segmentation combining active contour + kmeans 
    1. ./manuscript_segmentation.m
  • 2nd step: run bash script for the inpaint the damaged areas detected in step 1
    1. ./manuscript_inpainting.sh

Results:

The experiment with number XXX is stored in the folder “./results/paper_results/testXXX” folder, where

  1. - 101,102,... refers to images from Manuscript 1;
  2. - 201,202,... refers to images from Manuscript 2.

Example

In folder ./results/paper_resultS/test101 we store the results of experiment 01 for Manuscript 1:

  • input_orig101.png: the real crop of Manuscript 1;
  • input101.png: the preprocessing (smoothing/texture removal);
  • overlap_SUPER101.png: overlap between image crop and supervised pixel inputs in blue
    squares (just one pixel, dilated for display purposes);
  • overlap_SUPERwithCV101.png: overlap between active contour regions in yellow and
    supervised pixel inputs in blue squares (just one pixel, dilated for display purposes);
  • overlap_CV101png: overlap between image crop and active contour regions in yellow;
  • overlap101.png: final overlap between image and segmentation from kmeans+activecontour
  • mask101.png: final segmentation mask
  • masked101.png: input for TV inpainting
  • TVinpainted101.png: TV inpainting result
  • PATCHinpainted101_PxP: results of nonlocal exemplar-based inpainting result with patch
    of size P and TV inpainting as initialization

The content and images of this work are part of the ILLUMINATED: Manuscripts in the making project.
We refers to the project website for more information and the related copyright notice.

License

BSD 3-Clause License