Non-Destructive Estimation of Paper Fiber Using Macro Images: A Comparative Evaluation of Network Architectures and Patch Sizes for Patch-Based Classification
Over the years, research in the field of cultural heritage preservation and document analysis has exponentially grown. In this study, we propose an advanced approach for non-destructive estimation of paper fibers using macro images. Expanding on studies that implemented EfficientNet-B0, we explore t...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-11-01
|
| Series: | NDT |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2813-477X/2/4/30 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832587789082820608 |
|---|---|
| author | Naoki Kamiya Kosuke Ashino Yasuhiro Sakai Yexin Zhou Yoichi Ohyanagi Koji Shibazaki |
| author_facet | Naoki Kamiya Kosuke Ashino Yasuhiro Sakai Yexin Zhou Yoichi Ohyanagi Koji Shibazaki |
| author_sort | Naoki Kamiya |
| collection | DOAJ |
| description | Over the years, research in the field of cultural heritage preservation and document analysis has exponentially grown. In this study, we propose an advanced approach for non-destructive estimation of paper fibers using macro images. Expanding on studies that implemented EfficientNet-B0, we explore the effectiveness of six other deep learning networks, including DenseNet-201, DarkNet-53, Inception-v3, Xception, Inception-ResNet-v2, and NASNet-Large, in conjunction with enlarged patch sizes. We experimentally classified three types of paper fibers, namely, kozo, mitsumata, and gampi. During the experiments, patch sizes of 500, 750, and 1000 pixels were evaluated and their impact on classification accuracy was analyzed. The experiments demonstrated that Inception-ResNet-v2 with 1000-pixel patches achieved the highest patch classification accuracy of 82.7%, whereas Xception with 750-pixel patches exhibited the best macro-image-based fiber estimation performance at 84.9%. Additionally, we assessed the efficacy of the method for images containing text, observing consistent improvements in the case of larger patch sizes. However, limitations exist in background patch availability for text-heavy images. This comprehensive evaluation of network architectures and patch sizes can significantly advance the field of non-destructive paper analysis, offering valuable insights into future developments in historical document examination and conservation science. |
| format | Article |
| id | doaj-art-28642e2b284d441997d40898a910af68 |
| institution | Kabale University |
| issn | 2813-477X |
| language | English |
| publishDate | 2024-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | NDT |
| spelling | doaj-art-28642e2b284d441997d40898a910af682025-01-24T13:44:20ZengMDPI AGNDT2813-477X2024-11-012448750310.3390/ndt2040030Non-Destructive Estimation of Paper Fiber Using Macro Images: A Comparative Evaluation of Network Architectures and Patch Sizes for Patch-Based ClassificationNaoki Kamiya0Kosuke Ashino1Yasuhiro Sakai2Yexin Zhou3Yoichi Ohyanagi4Koji Shibazaki5School of Information Science and Technology, Aichi Prefectural University, Nagakute 480-1198, JapanGraduate School of Information Science and Technology, Aichi Prefectural University, Nagakute 480-1198, JapanSchool of Information Science and Technology, Aichi Prefectural University, Nagakute 480-1198, JapanLife and Culture Department, Seirei Women’s Junior College, Akita 011-0937, JapanFaculty of Fine Arts, Aichi University of the Arts, Nagakute 480-1194, JapanFaculty of Fine Arts, Aichi University of the Arts, Nagakute 480-1194, JapanOver the years, research in the field of cultural heritage preservation and document analysis has exponentially grown. In this study, we propose an advanced approach for non-destructive estimation of paper fibers using macro images. Expanding on studies that implemented EfficientNet-B0, we explore the effectiveness of six other deep learning networks, including DenseNet-201, DarkNet-53, Inception-v3, Xception, Inception-ResNet-v2, and NASNet-Large, in conjunction with enlarged patch sizes. We experimentally classified three types of paper fibers, namely, kozo, mitsumata, and gampi. During the experiments, patch sizes of 500, 750, and 1000 pixels were evaluated and their impact on classification accuracy was analyzed. The experiments demonstrated that Inception-ResNet-v2 with 1000-pixel patches achieved the highest patch classification accuracy of 82.7%, whereas Xception with 750-pixel patches exhibited the best macro-image-based fiber estimation performance at 84.9%. Additionally, we assessed the efficacy of the method for images containing text, observing consistent improvements in the case of larger patch sizes. However, limitations exist in background patch availability for text-heavy images. This comprehensive evaluation of network architectures and patch sizes can significantly advance the field of non-destructive paper analysis, offering valuable insights into future developments in historical document examination and conservation science.https://www.mdpi.com/2813-477X/2/4/30paper fiber estimationnon-destructive analysispatch-based image classificationdeep learning |
| spellingShingle | Naoki Kamiya Kosuke Ashino Yasuhiro Sakai Yexin Zhou Yoichi Ohyanagi Koji Shibazaki Non-Destructive Estimation of Paper Fiber Using Macro Images: A Comparative Evaluation of Network Architectures and Patch Sizes for Patch-Based Classification NDT paper fiber estimation non-destructive analysis patch-based image classification deep learning |
| title | Non-Destructive Estimation of Paper Fiber Using Macro Images: A Comparative Evaluation of Network Architectures and Patch Sizes for Patch-Based Classification |
| title_full | Non-Destructive Estimation of Paper Fiber Using Macro Images: A Comparative Evaluation of Network Architectures and Patch Sizes for Patch-Based Classification |
| title_fullStr | Non-Destructive Estimation of Paper Fiber Using Macro Images: A Comparative Evaluation of Network Architectures and Patch Sizes for Patch-Based Classification |
| title_full_unstemmed | Non-Destructive Estimation of Paper Fiber Using Macro Images: A Comparative Evaluation of Network Architectures and Patch Sizes for Patch-Based Classification |
| title_short | Non-Destructive Estimation of Paper Fiber Using Macro Images: A Comparative Evaluation of Network Architectures and Patch Sizes for Patch-Based Classification |
| title_sort | non destructive estimation of paper fiber using macro images a comparative evaluation of network architectures and patch sizes for patch based classification |
| topic | paper fiber estimation non-destructive analysis patch-based image classification deep learning |
| url | https://www.mdpi.com/2813-477X/2/4/30 |
| work_keys_str_mv | AT naokikamiya nondestructiveestimationofpaperfiberusingmacroimagesacomparativeevaluationofnetworkarchitecturesandpatchsizesforpatchbasedclassification AT kosukeashino nondestructiveestimationofpaperfiberusingmacroimagesacomparativeevaluationofnetworkarchitecturesandpatchsizesforpatchbasedclassification AT yasuhirosakai nondestructiveestimationofpaperfiberusingmacroimagesacomparativeevaluationofnetworkarchitecturesandpatchsizesforpatchbasedclassification AT yexinzhou nondestructiveestimationofpaperfiberusingmacroimagesacomparativeevaluationofnetworkarchitecturesandpatchsizesforpatchbasedclassification AT yoichiohyanagi nondestructiveestimationofpaperfiberusingmacroimagesacomparativeevaluationofnetworkarchitecturesandpatchsizesforpatchbasedclassification AT kojishibazaki nondestructiveestimationofpaperfiberusingmacroimagesacomparativeevaluationofnetworkarchitecturesandpatchsizesforpatchbasedclassification |