Single Document Image Highlight Removal via A Large-Scale Real-World Dataset and A Location-Aware Network

2025-04-19

AI Summary

To address the severe degradation of text readability in reflective document images under ambient illumination caused by specular highlights, this paper introduces DocHR14K—the first large-scale, high-resolution, real-scenario dataset of specular highlight-contaminated documents (14,902 image pairs)—and proposes L2HRNet, a network leveraging Highlight Location Prior (HLP). Methodologically, it pioneers a multi-category, multi-illumination real-world highlight document dataset; devises an unsupervised mechanism for highlight location prior estimation; and designs a Laplacian pyramid architecture integrating location-guided feature modulation and diffusion-based enhancement. On DocHR14K, L2HRNet achieves +5.01 dB PSNR gain and −13.17% RMSE reduction over prior methods. It also attains state-of-the-art performance across three public benchmarks, significantly advancing the practical deployment of specular highlight removal in document imaging.

Technology Category

• Computer Vision: Image and Video Retrieval
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

Reflective documents often suffer from specular highlights under ambient lighting, severely hindering text readability and degrading overall visual quality. Although recent deep learning methods show promise in highlight removal, they remain suboptimal for document images, primarily due to the lack of dedicated datasets and tailored architectural designs. To tackle these challenges, we present DocHR14K, a large-scale real-world dataset comprising 14,902 high-resolution image pairs across six document categories and various lighting conditions. To the best of our knowledge, this is the first high-resolution dataset for document highlight removal that captures a wide range of real-world lighting conditions. Additionally, motivated by the observation that the residual map between highlighted and clean images naturally reveals the spatial structure of highlight regions, we propose a simple yet effective Highlight Location Prior (HLP) to estimate highlight masks without human annotations. Building on this prior, we present the Location-Aware Laplacian Pyramid Highlight Removal Network (L2HRNet), which effectively removes highlights by leveraging estimated priors and incorporates diffusion module to restore details. Extensive experiments demonstrate that DocHR14K improves highlight removal under diverse lighting conditions. Our L2HRNet achieves state-of-the-art performance across three benchmark datasets, including a 5.01% increase in PSNR and a 13.17% reduction in RMSE on DocHR14K.

Problem

• Removing specular highlights from reflective document images
• Lack of dedicated datasets for document highlight removal
• Improving text readability and visual quality under various lighting

Innovation

• Large-scale real-world dataset DocHR14K
• Highlight Location Prior without annotations
• Location-Aware Laplacian Pyramid Network L2HRNet