LayeredDoc: Domain Adaptive Document Restoration with a Layer Separation Approach

TL;DR

LayeredDoc introduces a hierarchical document restoration method that separates text and graphic layers, enabling better domain adaptation and improved performance across diverse document types.

Contribution

The paper presents a novel layered approach for document image restoration that enhances domain adaptability by separating graphic and textual information.

Findings

Strong generalization on real-world datasets

Effective domain adaptation demonstrated

Hierarchical separation improves restoration quality

Abstract

The rapid evolution of intelligent document processing systems demands robust solutions that adapt to diverse domains without extensive retraining. Traditional methods often falter with variable document types, leading to poor performance. To overcome these limitations, this paper introduces a text-graphic layer separation approach that enhances domain adaptability in document image restoration (DIR) systems. We propose LayeredDoc, which utilizes two layers of information: the first targets coarse-grained graphic components, while the second refines machine-printed textual content. This hierarchical DIR framework dynamically adjusts to the characteristics of the input document, facilitating effective domain adaptation. We evaluated our approach both qualitatively and quantitatively using a new real-world dataset, LayeredDocDB, developed for this study. Initially trained on a synthetically generated dataset, our model demonstrates strong generalization capabilities for the DIR task, offering a promising solution for handling variability in real-world data. Our code is accessible on GitHub.