DvD: Unleashing a Generative Paradigm for Document Dewarping via Coordinates-based Diffusion Model

2025-05-28

AI Summary

Document dewarping aims to correct geometric distortions in photographed document images to enhance text readability and structural fidelity, yet existing methods struggle with preserving structural integrity. This paper proposes the Coordinate Mapping Diffusion Model (CMDM), which formulates distortion correction as a coordinate-level generative task. CMDM introduces a novel coordinate-level denoising mechanism to avoid pixel-level reconstruction artifacts; a time-varying conditional refinement module to improve structural consistency; and multi-scale feature fusion coupled with realistic distortion modeling. We further construct AnyPhotoDoc6300—the first large-scale, cross-domain benchmark of real-world distorted/corrected image pairs (6,300 pairs). Extensive experiments demonstrate that CMDM achieves state-of-the-art performance on DocUNet, DIR300, and AnyPhotoDoc6300, balancing accuracy and efficiency. Code and the benchmark will be publicly released.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Search and Optimization: Distributed Search

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Document dewarping aims to rectify deformations in photographic document images, thus improving text readability, which has attracted much attention and made great progress, but it is still challenging to preserve document structures. Given recent advances in diffusion models, it is natural for us to consider their potential applicability to document dewarping. However, it is far from straightforward to adopt diffusion models in document dewarping due to their unfaithful control on highly complex document images (e.g., 2000$ imes$3000 resolution). In this paper, we propose DvD, the first generative model to tackle document extbf{D}ewarping extbf{v}ia a extbf{D}iffusion framework. To be specific, DvD introduces a coordinate-level denoising instead of typical pixel-level denoising, generating a mapping for deformation rectification. In addition, we further propose a time-variant condition refinement mechanism to enhance the preservation of document structures. In experiments, we find that current document dewarping benchmarks can not evaluate dewarping models comprehensively. To this end, we present AnyPhotoDoc6300, a rigorously designed large-scale document dewarping benchmark comprising 6,300 real image pairs across three distinct domains, enabling fine-grained evaluation of dewarping models. Comprehensive experiments demonstrate that our proposed DvD can achieve state-of-the-art performance with acceptable computational efficiency on multiple metrics across various benchmarks including DocUNet, DIR300, and AnyPhotoDoc6300. The new benchmark and code will be publicly available.

Problem

• Rectify deformations in document images for better readability
• Preserve document structures during dewarping process
• Evaluate dewarping models comprehensively with new benchmark

Innovation

• Coordinates-based diffusion model for dewarping
• Time-variant condition refinement mechanism
• Large-scale benchmark AnyPhotoDoc6300 for evaluation