ChainLineNet: Deep-Learning-Based Segmentation and Parameterization of Chain Lines in Historical Prints
The paper structure of historical prints is sort of a unique fingerprint. Paper with the same origin shows similar chain line distances. As the manual measurement of chain line distances is time consuming, the automatic detection of chain lines is beneficial. We propose an end-to-end trainable deep...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-07-01
|
| Series: | Journal of Imaging |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2313-433X/7/7/120 |
| _version_ | 1797526830299742208 |
|---|---|
| author | Aline Sindel Thomas Klinke Andreas Maier Vincent Christlein |
| author_facet | Aline Sindel Thomas Klinke Andreas Maier Vincent Christlein |
| author_sort | Aline Sindel |
| collection | DOAJ |
| description | The paper structure of historical prints is sort of a unique fingerprint. Paper with the same origin shows similar chain line distances. As the manual measurement of chain line distances is time consuming, the automatic detection of chain lines is beneficial. We propose an end-to-end trainable deep learning method for segmentation and parameterization of chain lines in transmitted light images of German prints from the 16th Century. We trained a conditional generative adversarial network with a multitask loss for line segmentation and line parameterization. We formulated a fully differentiable pipeline for line coordinates’ estimation that consists of line segmentation, horizontal line alignment, and 2D Fourier filtering of line segments, line region proposals, and differentiable line fitting. We created a dataset of high-resolution transmitted light images of historical prints with manual line coordinate annotations. Our method shows superior qualitative and quantitative chain line detection results with high accuracy and reliability on our historical dataset in comparison to competing methods. Further, we demonstrated that our method achieves a low error of less than 0.7 mm in comparison to manually measured chain line distances. |
| first_indexed | 2024-03-10T09:35:53Z |
| format | Article |
| id | doaj.art-f9dee18369be449a98b2ef99ab9e434e |
| institution | Directory Open Access Journal |
| issn | 2313-433X |
| language | English |
| last_indexed | 2024-03-10T09:35:53Z |
| publishDate | 2021-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Journal of Imaging |
| spelling | doaj.art-f9dee18369be449a98b2ef99ab9e434e2023-11-22T04:08:47ZengMDPI AGJournal of Imaging2313-433X2021-07-017712010.3390/jimaging7070120ChainLineNet: Deep-Learning-Based Segmentation and Parameterization of Chain Lines in Historical PrintsAline Sindel0Thomas Klinke1Andreas Maier2Vincent Christlein3Pattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), 91058 Erlangen, GermanyCologne Institute of Conservation Sciences (CICS), Technische Hochschule Köln, 50678 Köln, GermanyPattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), 91058 Erlangen, GermanyPattern Recognition Lab, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), 91058 Erlangen, GermanyThe paper structure of historical prints is sort of a unique fingerprint. Paper with the same origin shows similar chain line distances. As the manual measurement of chain line distances is time consuming, the automatic detection of chain lines is beneficial. We propose an end-to-end trainable deep learning method for segmentation and parameterization of chain lines in transmitted light images of German prints from the 16th Century. We trained a conditional generative adversarial network with a multitask loss for line segmentation and line parameterization. We formulated a fully differentiable pipeline for line coordinates’ estimation that consists of line segmentation, horizontal line alignment, and 2D Fourier filtering of line segments, line region proposals, and differentiable line fitting. We created a dataset of high-resolution transmitted light images of historical prints with manual line coordinate annotations. Our method shows superior qualitative and quantitative chain line detection results with high accuracy and reliability on our historical dataset in comparison to competing methods. Further, we demonstrated that our method achieves a low error of less than 0.7 mm in comparison to manually measured chain line distances.https://www.mdpi.com/2313-433X/7/7/120line segmentationline detectionline parameterizationgenerative adversarial networksFourier transformdifferentiable line fitting |
| spellingShingle | Aline Sindel Thomas Klinke Andreas Maier Vincent Christlein ChainLineNet: Deep-Learning-Based Segmentation and Parameterization of Chain Lines in Historical Prints Journal of Imaging line segmentation line detection line parameterization generative adversarial networks Fourier transform differentiable line fitting |
| title | ChainLineNet: Deep-Learning-Based Segmentation and Parameterization of Chain Lines in Historical Prints |
| title_full | ChainLineNet: Deep-Learning-Based Segmentation and Parameterization of Chain Lines in Historical Prints |
| title_fullStr | ChainLineNet: Deep-Learning-Based Segmentation and Parameterization of Chain Lines in Historical Prints |
| title_full_unstemmed | ChainLineNet: Deep-Learning-Based Segmentation and Parameterization of Chain Lines in Historical Prints |
| title_short | ChainLineNet: Deep-Learning-Based Segmentation and Parameterization of Chain Lines in Historical Prints |
| title_sort | chainlinenet deep learning based segmentation and parameterization of chain lines in historical prints |
| topic | line segmentation line detection line parameterization generative adversarial networks Fourier transform differentiable line fitting |
| url | https://www.mdpi.com/2313-433X/7/7/120 |
| work_keys_str_mv | AT alinesindel chainlinenetdeeplearningbasedsegmentationandparameterizationofchainlinesinhistoricalprints AT thomasklinke chainlinenetdeeplearningbasedsegmentationandparameterizationofchainlinesinhistoricalprints AT andreasmaier chainlinenetdeeplearningbasedsegmentationandparameterizationofchainlinesinhistoricalprints AT vincentchristlein chainlinenetdeeplearningbasedsegmentationandparameterizationofchainlinesinhistoricalprints |