Simulation-Aided Infrared Thermography with Faster R-CNN-ECA Model and LRTDTV Denoising Method: the Case-Study of Ancient Polyptychs

EasyChair Preprint 15209

Abstract

In this study, we investigate how to automatically and efficiently detect defects in ancient polyptychs by infrared thermography, combined with numerical simulation, deep learning networks and machine learning algorithms. Through an innovative improved Faster R-CNN model and LRTDTV denoising method, the recognition of surface and internal defects of ancient artworks is effectively improved. This enhanced Faster R-CNN model incorporates an effective channel attention mechanism in the feature extraction stage, significantly boosting the model's performance in recognizing small defects. Comparisons with the original Faster R-CNN model show that the average precision at an intersection over union of 0.5 has increased to 87.3% for the improved model. Notably, the average precision for detecting small defects has risen to 54.8%. The experimental results verify the practicality and efficiency of the method in cultural heritage conservation, which helps to maximize the conservation 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 objects while avoiding secondary damage to the samples, providing effective technical support for cultural heritage conservation.

Keyphrases: Attention Mechanism, Faster R-CNN network, Infrared thermography, Machine Learning Algorithms, cultural heritage, deep learning, defect detection, numerical simulation

@booklet{EasyChair:15209,
  author    = {Xin Wang and Guimin Jiang and Miranda Mostacci and Stefano Sfarra and Hai Zhang},
  title     = {Simulation-Aided Infrared Thermography with Faster R-CNN-ECA Model and LRTDTV Denoising Method: the Case-Study of Ancient Polyptychs},
  howpublished = {EasyChair Preprint 15209},
  year      = {EasyChair, 2024}}