MESA: A Training-Free Multi-Exemplar Deep Framework for Restoring Ancient Inscription Textures

2026-04-19

AI Summary

This study addresses the challenge of missing text in ancient inscriptions caused by weathering and damage, which hinders legibility and scholarly analysis. The authors propose a training-free, multi-exemplar deep inpainting method that leverages intact regions from the same stele or visually similar characters as exemplars. Texture and stroke features are extracted using Gram matrices derived from VGG19, and the optimal exemplar is selected layer-wise via minimum mean squared displacement (MSD). To enhance structural fidelity, character widths estimated by OCR guide adaptive layer weighting, while a trained mask constrains the synthesized regions. This approach uniquely integrates multi-exemplar guidance, character-width-aware weighting, and mask-based synthesis, significantly outperforming existing single-image inpainting, GAN-based, and traditional restoration methods in both stylistic consistency and reconstruction accuracy, and offers practical guidelines for real-world restoration tasks.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Ancient inscriptions frequently suffer missing or corrupted regions from fragmentation, erosion, or other damage, hindering reading, and analysis. We review prior image restoration methods and their applicability to inscription image recovery, then introduce MESA (Multi-Exemplar, Style-Aware) -an image-level restoration method that uses well-preserved exemplar inscriptions (from the same epigraphic monument, material, or similar letterforms) to guide reconstruction of damaged text. MESA encodes VGG19 convolutional features as Gram matrices to capture exemplar texture, style, and stroke structure; for each neural network layer it selects the exemplar minimizing Mean-Squared Displacement (MSD) to the damaged input. Layer-wise contribution weights are derived from Optical Character Recognition-estimated character widths in the exemplar set to bias filters toward scales matching letter geometry, and a training mask preserves intact regions so synthesis is restricted to damaged areas. We also summarize prior network architectures and exemplar and single-image synthesis, inpainting, and Generative Adversarial Network (GAN) approaches, highlighting limitations that MESA addresses. Comparative experiments demonstrate the advantages of MESA. Finally, we provide a practical roadmap for choosing restoration strategies given available exemplars and metadata.

Problem

• ancient inscriptions
• image restoration
• missing regions
• texture reconstruction
• damaged text

Innovation

• multi-exemplar
• style-aware
• training-free
• Gram matrix
• inscription restoration