ABCDEFGHIJKLMNOPQRSTUVWXYZAAABAC
1
Title of the articlePublication YearResearch questionHypothesisMethodologyData collectionData analysisResultsConclusionLimitationsFuture researchSample sizeData source
2
Innovative uses of 3D digital technologies to assist the restoration of a fragmented terracotta statue2013How can innovative computer-based technologies help Cultural Heritage restorers in finding proper recombination hypotheses of fragmented artwork, reassembling it, studying hypotheses of the original painted decoration, documenting the restoration process and presenting it to the public?The level of contribution that innovative computer-based technologies could bring to cultural heritage restorers is extremely important.The authors designed and implemented innovative methodologies tailored to support and assist high-quality 3D digital models in the framework of a complex restoration case using computer-aided technologies.through 3D scanning and renderingusing 3D digitization and graphics technologiesThe 3D digital models and 3D printing techniques were employed as a significant contribution to restoring the terracotta statueInnovative 3D digital technologies have a great potential to assist the cultural heritage restorers in solving complex restoration problems in finding hypotheses, reassembling the artwork, studying the decoration, documenting, and presenting the restored artwork to the public.The high cost of software and hardware, including 3D digitization equipment and graphics software, limit the use of 3D digitization technology to significant cultural heritage restorations.Future research could be focused on exploring newer technologies and tools to further enhance and utilize virtual reconstruction, virtual restoration, and documentation techniques to support the restoration of culturally significant artworks that are fragmented or damaged.Not mentionedThe primary data source is the restoration of the terracotta statue in this particular case study.
3
Virtual Reconstruction and Representation of an Archaeological Terracotta Statue2015How can 3D technology be applied in the field of heritage restoration and, in particular, for the virtual reconstruction of an archaeological terracotta statue by inferring the topology of the missing parts?Using 3D technology to assist the restoration process in both the reassembly process of the fragments and the virtual reconstruction and visualization of the complete sculpture.A combination of triangulation laser scanning and photogrammetry techniques to acquire the geometry and diffuse texture of the statue fragments. They then applied an automatic reassembly technique to the digital models of the fragments to consider the best alignments between themselves (without the need to manipulate the original ones) and lead the reassembly and adhesion processes of the fragments. Finally, they used immersive 3D interactive visualization to present the virtual reconstruction of the entire statue. Acquired the geometry of the model using a triangulation laser scanner and generated the diffuse texture by means of photogrammetry techniques using a Nikon D70 camera.Used an automatic reassembly technique to operate on digital models by considering the best alignments between fragments without the need to manipulate the original ones. Successfully created a virtual reconstruction of a late 15th-century terracotta statue using 3D technology and an automatic reassembly technique.3D technology offers a fully automated solution for the restoration and exhibition of archaeological artifacts, particularly when it comes to the virtual reconstruction of missing fragments. Proposed technique works best only with unconstrained reassembly problems with no assumptions on fragments topology.Gestural interfaces can be used to assemble/disassemble parts of the 3D model with the potential offered by Oculus Rift and the stereoscopic visionSample is not usedFrom the archaeological terracotta statue "Crist del Fossar" from the collection of the Museu de Prehistòria de València, which was found severely damaged and partially conserved as the arms of the figure were missing.
4
Digital Restoration Using Image-Based 3D Models2015Digital restoration using digital 3D models exclusively produced by means of image-based technologiesNot mentionedUsed an image-based approach to produce 3D models of archaeological finds, and carried out a comparative test on different image datasets with different ISO and different resolution, to provide a metric comparison of results, as an operational base.The photographic survey has concerned a fragment of a loricata statue of emperor, from the frons scenae of the roman theatre of Lecce, actually exposed at the City Historic MuseumCompared 3D models obtained with photos with ISO values and different resolution of images. The resulting models were imported into a 3D modelling software, and an element of 15 cm per side was isolated in order to compare the subdivision of the polygons and the distribution of points.Digital restoration of sculptural elements gets significant benefits with the use of 3D models.The full 3D approach to the problem of the restoration of archaeological finds, extremely damaged, raises several points of interest, including sculpting and painting directly on surfaces, as well with the possibility of reintegration of the missing parts using other sculptural elements in similar subjects best preserved.Not mentionedNot MentionedNot mentionedPhotographic survey of a fragment of a loricata statue of emperor, from the frons scenae of the roman theatre of Lecce, actually exposed at the City Historic Museum
5
Computer Aided Restoration Tools to Assist The Conservation of an Ancient Sculpture. The Colossal Statue of Zeus Enthroned2017To provide effective geometrical-formal investigation tools in the frame of the conservation work Use of Computer Aided Restoration digital procedures could guide the integration method of an artifact, innovating and implementing the traditional investigation methods to assist the conservation of the legs of the throne, especially the integration of the missing part.Use of 3D modeling, virtual recomposition, 3D printingThe campaign of indirect detection was carried out by the team in 5 working days, with the collaboration of the conservator for the handling of the pieces. The geometric survey helps in evaluating the state of material preservation of the external portions of the objectThe high reliability of this tool guarantees obtaining a copy with the same proportion of the original one, not easy with traditional casting techniques. The exact model obtained by printing, it can be made with different materials chosen on the basis of the environmental condition 3D digital technologies is a useful tool for the conservation of artifactsThe processing of the surfaces takes a long time in relation to the surface detail and the size of the scanning object. The application, in sequence of the acquisition data, digital modeling, and 3D printing, permits to elaborate hypothesis and suggestions, hardly obtainable with other applications, enhancing the field of innovative 3D applications from the philological to the virtual conservation.Not mentionedStatue enthroned of the god Zeus from Soluntum, second half of the XIX century (Archeological Museum of Palermo A. Salinas).
6
3D Reconstruction of Incomplete Archaeological Objects Using a Generative Adversarial Network2018How to employ a data-driven approach to repair and conserve archaeological objects by using an object reconstruction generative adversarial network.A generative adversarial network (GAN) architecture, in combination with two loss objectives (a completion loss and an improved Wasserstein GAN loss), can effectively predict the missing geometry of damaged objects.The method combines an encoder-decoder 3D deep neural network with two loss objectives: a completion loss and an Improved Wasserstein GAN loss.Uses 3D scans of incomplete archaeological objects for its data.Variety of evaluation metrics to assess the performance of the method, including symmetric mean absolute percentage error (SMAPE), root mean squared error (RMSE), and Intersection over Union (IoU).Proposed method could recover most of the information from damaged objects, even in cases where more than half of the voxels were missing, without producing many errors.A generative adversarial network (GAN) architecture, in combination with two loss objectives, can effectively predict the missing geometry of damaged archaeological objects. The method assumes that man-made objects exhibit some kind of structure and regularity, which may not always be the case with archaeological objects.Explore the use of different GAN architectures and loss functions to improve the used method. Also, the method could be extended to other domains, such as medical imaging.Not mentioned3D scans of incomplete archaeological objects
7
Recognition Techniques in Buddhist Iconography and Challenges2018Successful application of pattern recognition techniques in the field of iconography has recently received significant attention.Can pattern recognition techniques be effectively applied in the field of Buddhist iconography to identify period of origin and age of ancient sculptures and artefacts? a critical analysis of the research that has been carried out in the field of Buddhist Iconography as well as the various techniques applied.The various research papers and studies carried out in the field of Buddhist iconography were used for analysis.A critical review and analysis of the existing literature, focusing on the various pattern recognition techniques applied in the field of Buddhist Iconography.The application of pattern recognition techniques has shown promise in identifying the period of origin and age of ancient sculptures and artefacts in the field of Buddhist Iconography.successful application of pattern recognition techniques in the field of Buddhist iconography is possible collection of Buddhist databases and its variant modes of capturing.enlarging the database by including Buddhist sculptures of different styles so that the intelligence of the system may be enhanced.Not mentionedVarious research papers
8
The Use of New Technologies in the Restoration and Conservation of Built Cultural Heritage / The Case of the Statue of Fouara, Setif, Algeria2019The potential of laser scanning technology in the restoration and conservation of built cultural heritageThe original laser scanning survey of the statue will provide helpful and accurate information to support its restoration, specifically the missing parts, the face and the breast.Used laser scanning technology to create a digital model of the statue before and after the vandalism incident. Then used the software to compare the two models and identify the damaged parts. Finally, used the data to create 2D section plans for use in the restoration process.Using terrestrial laser scanning technologyUsed software to compare the pre- and post- vandalism digital models and generate 2D section plans.Able to use the data to create a digital model of the statue and generate 2D section plans for use in the restoration process.Laser scanning technology could be a useful tool in the restoration and conservation of built cultural heritage.Primarily focuses on one case study and does not explore the broader applicability of laser scanning technologyAutomated procedures and additional data that can be obtained from laser scanning technology for use in restoration and conservation efforts.1The statue itself
9
Historical and Modern Features for Buddha Statue Classification2019How do the historical construction guidelines of Buddha statues reflect in a dataset of Buddha statues? Can deep learning techniques be used to classify different types of Buddha statues?The study of the historical construction guidelines of Buddha statues and their proportions will reflect in a dataset of Buddha statues and can be used to classify different types of Buddha statues. Deep learning techniques can also help classify the different types of Buddha statues.The study used a mixed-methods approach to investigate the features of Buddha statues. The researchers used an automatic landmark detection algorithm to recover the construction guidelines of Buddha faces. They further used iconometry (the study of art pieces measurements) to investigate the differences between the classes of Buddha statues. Additionally, they employed deep learning algorithms to investigate different deep features and classification tasks of the Buddha statues.Collaborated with experts to investigate three important styles of Buddha statues: ancient Chinese statues spreading between the IV and XIII centuries, Japanese statues during the Heian period (794-1185), and Japanese statues during the Kamakura era (1185-1333). The researchers analyzed the different data features by using deep learning methods for classification tasks such as style classification, statue type classification, dimension classification, century classification, and material classification. the construction guidelines of Buddha statues reflect in the dataset of Buddha statues. The landmark detection algorithm accurately detected the iconometric proportions of different facial regions.the historical construction guidelines of Buddha statues can be recovered from photographs to identify the iconometric proportions of different facial regions.proposed landmark detection algorithm did not work on some pictures, which might decrease the overall accuracy of the analysis.developing a specific landmark detector for Buddha statues to improve the landmark analysis accuracy.1393The data came from 4 series of books that the experts scanned resulting in 6811 scanned images
10
Generative Adversarial Networks for Single Photo 3D Reconstruction2019Single photo 3D reconstruction problem for lost cultural objects for which only a few images are remaining.The network can generate voxel models of previously unseen objects using object silhouettes present on the input image and the knowledge obtained during a training stage.Generative Adversarial Network (GAN) based on image-to-voxel translation network (Z-GAN) as a starting point. The Z-GAN network utilizes the skip connections in the generator network to transfer 2D features to a 3D voxel model. Using heritage datasetsQuantitative and Qualitative analyses of the results were doneProposed method can produce voxel-based models for complex structures and lost heritage still available in crowdsourced imagesProposed method shows promising results in the 3D reconstruction of cultural heritage scenes from a single image. Requires a large number of images for training. The method can also be used in other domains such as robotics and autonomous vehicles.600 images for training their Z-GAN network out of 680 images of the Neptune temple (Paestum, Italy).Images of the Neptune temple and Cerere temple
11
Noncontact restoration of missing parts of stone Buddha statue based on three-dimensional virtual modeling and assembly simulation2020Can 3D scanning, virtual restoration modeling, and 3D printing be used as a noncontact approach to restore a damaged stone-seated Bodhisattva (stone Buddha statue)?3D scanning, virtual restoration modeling, and 3D printing can be used as a non-contact approach for restoring a damaged stone-seated Bodhisattva (stone Buddha statue).3D scanning, virtual restoration modeling, and 3D printing were used as a non-contact approach for restoring a damaged stone-seated Bodhisattva (stone Buddha statue).3D scanning of the physically damaged stone Buddha statue and digital virtual restoration of the missing parts using a haptic modeling system.3D model with an average point density of 0.2 mm was created by integrating the fixed high-precision scanningThe study showed that 3D scanning, virtual restoration modeling, and 3D printing can be used as a non-contact approach for restoring a damaged stone-seated Bodhisattva (stone Buddha statue).modern technologies and materials can be used for the restoration, education, and exhibition of cultural artifacts.limited to a single stone Buddha statue.explore the use of algorithms to modify the modeling of the interference between fractured surfaces to reduce the frequency of design mockup assembly.1 The data was collected through 3D scanning of the physically damaged stone Buddha statue and virtual restoration modeling of the missing parts using a haptic modeling system.
12
ARQGAN: An evaluation of generative adversarial network approaches for automatic virtual inpainting restoration of Greek temples2021Can generative adversarial networks be used for virtual inpainting restoration of artificial landscape images containing archaeological remains of Greek temples?The use of generative adversarial networks will provide effective results for virtual inpainting restoration of artificial landscape images containing archaeological remains of Greek temples.using generative adversarial networks for virtual inpainting restoration, segmented training, mathematical metrics evaluation, and surveys of students and professionals.The data used in this research consisted of artificial landscape images containing archaeological remains of Greek temples.using mathematical metrics, student and professional surveys, and boxplots of evaluation scores.Both mathematical metrics evaluation and surveys showed good results.The use of generative adversarial networks for virtual inpainting restoration of artificial landscape images containing archaeological remains of Greek temples is effective.subjectivity of personal opinion in the evaluation of the model's performance.application of this method to other architectural styles and the use of other metrics for evaluation beyond personal opinion.1The data used in this research were obtained from artificial landscape images containing archaeological remains of Greek temples.
13
Built Year Prediction from Buddha Face with Heterogeneous Labels2021 how to automatically estimate the built years of Buddha statues based only on their face images.a neural network model can be built that accurately estimates the built years of Buddha statues based on their face images, even when labels are missing or estimated imprecisely. building a deep learning-based model that uses a loss function consisting of three terms: an MSE loss for built year estimation, a KL divergence-based loss for handling samples with both exact built year and possible range of built years, and a regularisation to utilise both labelled and unlabelled samples based on manifold assumption. These three terms are combined in the training process.The research used the dataset of Buddha statues presented by Renoust et al. (2019), which is comprised of scanned images of Buddha faces. This dataset provides a rich set of annotations on built time, materials, etc. trained a regression model to compute the built year from the image embedding extracted with a convolutional neural network.the proposed method was able to estimate built years for given images with 37.5 years of mean absolute error on the test set.proposed method outperformed state-of-the-art methods and baselines by a significant margin in estimating built years of Buddha statues based on face images.the dataset used is small and limited in the number of labels that have been provided. Future research can incorporate additional Buddha statues related information available in the dataset that can be correlated with the built time, such as built material and original location, and handle highly heterogeneous data as the dataset has a lot of missing entries.1340The research used the dataset of Buddha statues presented by Renoust et al. (2019).
14
Restoration of artwork using deep neural networks2021Can deep neural networks be used for virtual restoration of the digitized artworks?Deep neural networks can be used effectively for virtual restoration of the digitized artworks.The proposed method is based on deep neural networks, which employs automatic mask generation based on Mask R-CNN and image inpainting using U-Net architecture with partial convolutions and automatic mask update.The sample images were obtained from the ArtImages dataset available on Kaggle.The approach is evaluated qualitatively as well as quantitatively using mean square error (MSE) and structural similarity index (SSIM) metrics.The results obtained show that the proposed approach is quite effective in virtual restoration of the digitized artworks.The digitally restored images can help in preserving heritage and culture.Not mentionedThe proposed approach can be extended to restore other types of artworks such as sculptures, engraving, drawings etc.1042The sample images were obtained from the ArtImages dataset available on Kaggle.
15
Reconstruction of Iberian ceramic potteries using generative adversarial networks2022 Can generative adversarial networks be used to reconstruct fragmented Iberian ceramic potteries?a customized Generative Adversarial Network (GAN), called IberianGAN, could reconstruct pottery profiles from partial fragments of an original piece. a customized Generative Adversarial Network (GAN), called IberianGAN, to reconstruct pottery profiles from partial fragments of an original piece.used a database of Iberian wheel-made pottery profiles from archaeological sites located in the upper valley of the Guadalquivir River in Spain. The database contained both complete and fragmented references.used quantitative and qualitative assessments to measure the quality of the reconstructed samples, along with domain expert evaluation with archaeologists.The results showed that IberianGAN was capable of generating potteries that satisfied the image, pottery morphometric structure, and expert validation criteria.proposed framework is a possible way to facilitate pottery reconstruction from partial fragments of an original piece.the network was trained always using a base or rim fragment, meaning that the model will always position a fragment as a base or rim. Furthermore, their approach uses large fragments during training and evaluation, and additional studies are needed to determine the minimum accepted size of a fragment for the model to perform as expectedproposed framework could be used beyond just ceramic pottery to reconstruct other archaeological (e.g. projectile points, historical buildings, etc.) and anthropological remains (e.g. crania, postcranial bones, etc.).1072The data came from a database of Iberian wheel-made pottery profiles belonging to archaeological sites located in the upper valley of the Guadalquivir River in Spain.
16
Intelligent Restoration Technology of Mural Digital Image Based on Machine Learning Algorithm2022how to implement an intelligent fresco digital image restoration technique based on machine learning algorithms to solve the fresco image restoration of conversion and structural decomposition defects.using a machine learning algorithm would improve the accuracy of classifying the color and structure of a damaged mural image while reducing the cost of restoration.deep learning model applied in combination with mural imagery scanning program.acquiring digital image information of murals with a scanning program.using the mean filter template to restore the color of mural digital image, and the Gaussian template to restore the color of image details. this method can effectively repair the local fuzzy features of mural digital images, with a restoration accuracy of more than 95.7%, and the image quality being good.using a machine learning algorithm in a deep learning model for mural image restoration effectively improved the image restoration accuracy and preserved the consistency of mural digital image structures.One limitation mentioned was that the restoration technology was not suitable for other types of damaged murals.Method can be used in other restoration tasks1The data source in this study was the Mogao cave mural database.
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100