Dual Dimensions Geometric Representation Learning Based Document Dewarping

2025-07-11

AI Summary

Document image dewarping remains challenged by insufficient modeling of curved distortions, as existing methods typically model only horizontal text lines while neglecting vertical structural constraints. To address this, we propose D2Dewarp—a novel deep network enabling joint geometric modeling of both horizontal and vertical text lines for the first time. Its key contributions include: (1) a cross-dimensional feature fusion module leveraging X-Y coordinates to enhance bidirectional structural interaction; (2) an automatic fine-grained annotation method to construct a large-scale distorted document dataset; and (3) an end-to-end architecture integrating text-line awareness and coordinate alignment, trained on synthetically generated data. Evaluated on public Chinese and English benchmarks, D2Dewarp achieves significant improvements over state-of-the-art methods in quantitative metrics (e.g., PSNR, SSIM) and delivers superior visual correction quality.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Neural Architectures and Foundation Models
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

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 image dewarping remains a challenging task in the deep learning era. While existing methods have improved by leveraging text line awareness, they typically focus only on a single horizontal dimension. In this paper, we propose a fine-grained deformation perception model that focuses on Dual Dimensions of document horizontal-vertical-lines to improve document Dewarping called D2Dewarp. It can perceive distortion trends in different directions across document details. To combine the horizontal and vertical granularity features, an effective fusion module based on X and Y coordinate is designed to facilitate interaction and constraint between the two dimensions for feature complementarity. Due to the lack of annotated line features in current public dewarping datasets, we also propose an automatic fine-grained annotation method using public document texture images and an automatic rendering engine to build a new large-scale distortion training dataset. The code and dataset will be publicly released. On public Chinese and English benchmarks, both quantitative and qualitative results show that our method achieves better rectification results compared with the state-of-the-art methods. The dataset will be publicly available at https://github.com/xiaomore/DocDewarpHV

Problem

• Improving document dewarping using dual horizontal-vertical line dimensions
• Addressing lack of annotated line features in dewarping datasets
• Achieving better rectification results on public benchmarks

Innovation

• Dual Dimensions model for horizontal-vertical lines
• Fusion module for X-Y coordinate feature interaction
• Automatic annotation for large-scale distortion dataset