In this study, we investigate how to automatically and efficiently detect defects in ancient polyptychs by infrared thermography (IRT), combined with numerical simulation, deep learning networks and machine learning algorithms. Through an innovative improved Faster-RCNN model and LRTDTV denoising method, the recognition of surface and internal defects of ancient artworks is effectively improved. This improved Faster-RCNN model introduces an effective channel attention (ECA) mechanism in the feature extraction stage, which significantly improves the performance of the model in recognizing small defects, and a comparison with the original Faster-RCNN model reveals that the average detection accuracy_50 (AP_50) of the improved model is significantly improved to 86.9%. The average precision_small (AP_s) especially improved to 59.1% when detecting small-size defects. The experimental results verify the practicality and efficiency of the method in cultural heritage protection, which helps to maximize the protection and transmission of cultural heritage. In addition, the method in this study can achieve fast and accurate detection of defects in any type of cultural heritage object while avoiding secondary damage to the samples, providing effective technical support for cultural heritage protection.

Abstract In recent years, the preservation and conservation of ancient cultural heritage necessitate the advancement of sophisticated non-destructive testing methodologies to minimize potential damage to artworks. Therefore, this study aims to develop an advanced method for detecting defects in ancient polyptychs using infrared thermography. The test subjects are two polyptych samples replicating a 14th-century artwork by Pietro Lorenzetti (1280/85–1348) with varied pigments and artificially induced defects. To address these challenges, an automatic defect detection model is proposed, integrating numerical simulation and image processing within the Faster R-CNN architecture, utilizing VGG16 as the backbone network for feature extraction. Meanwhile, the model innovatively incorporates the efficient channel attention mechanism after the feature extraction stage, which significantly improves the feature characterization performance of the model in identifying small defects in ancient polyptychs. During training, numerical simulation is utilized to augment the infrared thermal image dataset, ensuring the accuracy of subsequent experimental sample testing. Empirical results demonstrate a substantial improvement in detection performance, compared with the original Faster R-CNN model, with the average precision at the intersection over union = 0.5 increasing to 87.3% and the average precision for small objects improving to 54.8%. These results highlight the practicality and effectiveness of the model, marking a significant progress in defect detection capability, providing a strong technical guarantee for the continuous conservation of cultural heritage, and offering directions for future studies.

In recent years, the conservation and protection of ancient cultural heritage have received increasing attention, and non-destructive testing (NDT), which can minimize the damage done to the test subject, plays an integral role therein. For instance, NDT through active infrared thermal imaging can be applied to ancient polyptychs, which can realize accurate detection of damage and defects existing on the surface and interior of the polyptychs. In this study, infrared thermography is used for non-invasive investigation and evaluation of two polyptych samples with different pigments and artificial defects, but both reproduced based on a painting by Pietro Lorenzetti (1280/85–1348) using the typical tempera technique of the century. It is noted that, to avoid as far as possible secondary damages done to the ancient cultural heritages, repeated damage-detection experiments are rarely carried out on the test subjects. To that end, numerical simulation is used to reveal the heat transfer properties and temperature distributions, as to perform procedural verification and reduce the number of experiments that need to be conducted on actual samples. Technique-wise, to improve the observability of the experimental results, a total variation regularized low-rank tensor decomposition algorithm is implemented to reduce the background noise and improve the contrast of the images. Furthermore, the efficacy of image processing is quantified through the structural-similarity evaluation.

Infrared thermography is a cost-effective non-destructive evaluation technique that plays a critical role in extracting information about defects in cultural heritage such as works of art. However, in-depth studies on the internal structure of contemporary artworks using infrared thermography remain lacking, and therefore, a deep convolutional autoencoder thermography (DCAT) data analysis method is proposed for defect detection of contemporary artworks. In the proposed method, original data are enhanced by convolution to reduce the noise and inhomogeneous background in the original thermal images; subsequently, a deep autoencoder is used to extract nonlinear features from the enhanced thermographic data, and the results of the hidden layer are visualised. These visualised images highlight information regarding the internal structure and defects of the artwork. The case study on a handmade replica of Picasso's 'La Bouteille de Suze', reproduced exactly as the original, has shown that DCAT significantly improved the accuracy of defect detection when compared with that of commonly used methods.

Early defects in artworks, which are unnoticeable by the naked eye, can evolve in severity if left unattended; therefore, their timely and accurate detection is of vital importance. Infrared-based non-destructive testing (NDT) methods are widely seen as an effective means to detect defects in artworks. However, most existing methods still rely on the human-based judgment of defects after processing the thermal images, which would inevitably lead to opinion bias and inconsistency. Deep-learning algorithms for computer vision are thought adequate to address the situation. On this point, an improved YOLOX algorithm, with the convolutional block attention module (CBAM) mechanism added to the network, is employed to improve the status quo. The method works with thermal imaging data under flash lamps for tempera paintings on canvas. A part of the famous "The Birth of Venus" by Botticelli ( & SIM;1485 ca.) was re-produced by a professional restorer, and three different defects were artificially introduced. To overcome the challenge pertaining to having too few defect samples, the dataset is expanded using finite-element simulation methods. The simulated data are used for training, whereas the replica is used for testing. Experimental results show that adding the CBAM mechanism after upsampling and downsampling of the path aggregation network can maximize network performance. The performance of the improved network herein proposed outperforms the benchmarks, with an average accuracy of 96.67%, a detection speed of 20.0 fps, and a total of 5.08 million parameters.& COPY; 2023 Consiglio Nazionale delle Ricerche (CNR). Published by Elsevier Masson SAS. All rights reserved.

With increasing attention paid to the protection of cultural relics, non-destructive testing (NDT) technologies are thought to be profoundly rewarding, resulting in a widespread uptake of feature-extraction algorithms and defect-detection techniques. Among various alternatives, infrared thermography (IRT) and Terahertz time-domain spectroscopy (THz-TDS) are non-invasive in nature and thus are appropriate for applications involving ancient buildings and artworks. The present study is motivated by the fact that online/offline background segmentation algorithms based on the Gaussian mixture model and widely used in video processing to distinguish between foreground and background, can be successfully integrated as a feature extraction tool with NDT. Since IRT and THz-TDS image sequences resemble a video, the image length and width can be taken as the first two dimensions, and time can serve as the third one. Such sequences can be processed effectively by the background segmentation algorithms, thus can help detect defects of different types at varying depths. The experimental section of the paper considers a tempera painting (a replica of Botticelli’s “The Birth of Venus”) with artificiallyintroduced defects. For benchmarking purposes, the background segmentation algorithms (based on a mixture of Gaussian models) are compared with the fast Fourier transform and principal component analysis to demonstrate the superior performance of the proposed novel algorithms.

The use of different spectral bands in the inspection of artworks is highly recommended to identify the maximum number of defects/anomalies (i.e., the targets), whose presence ought to be known before any possible restoration action. Although an artwork cannot be considered as a composite material in which the zero-defect theory is usually followed by scientists, it is possible to state that the preservation of a multi-layered structure fabricated by the artist's hands is based on a methodological analysis, where the use of non-destructive testing methods is highly desirable. In this paper, the infrared thermography and hyperspectral imaging methods were applied to identify both fabricated and non-fabricated targets in a canvas painting mocking up the famous character "Venus" by Botticelli. The pulse-compression thermography technique was used to retrieve info about the inner structure of the sample and low power light-emitting diode (LED) chips, whose emission was modulated via a pseudo-noise sequence, were exploited as the heat source for minimizing the heat radiated on the sample surface. Hyper-spectral imaging was employed to detect surface and subsurface features such as pentimenti and facial contours. The results demonstrate how the application of statistical algorithms (i.e., principal component and independent component analyses) maximized the number of targets retrieved during the post-acquisition steps for both the employed techniques. Finally, the best results obtained by both techniques and post-processing methods were fused together, resulting in a clear targets map, in which both the surface, subsurface and deeper information are all shown at a glance.