用Vision Mamba提升文档二值化质量，尤其适合历史文档
Adaptive Multi Scale Document Binarisation Using Vision Mamba
- 基于Vision Mamba构建新架构，线性扩展处理长序列
- 引入DoG特征增强多尺度高频信息，输出更清晰细节
- 适合需保留细粒度文字的古籍/历史文档分析
提升文档图像（尤其是历史文档）的可读性对文档图像分析至关重要。现有方法包括基于卷积、基于Transformer及混合卷积-Transformer架构。尽管混合模型弥补了纯卷积或纯Transformer的局限，但常面临二次时间复杂度问题。本文提出一种基于Mamba的文档二值化架构，通过线性扩展高效处理长序列并优化内存使用。此外，我们创新性地在跳跃连接中引入差分高斯（DoG）特征，借鉴传统信号处理技术。这些多尺度高频特征使模型能生成高质量、细节丰富的输出。
原文摘要 · Abstract (English)
Enhancing and preserving the readability of document images, particularly historical ones, is crucial for effective document image analysis. Numerous models have been proposed for this task, including convolutional-based, transformer-based, and hybrid convolutional-transformer architectures. While hybrid models address the limitations of purely convolutional or transformer-based methods, they often suffer from issues like quadratic time complexity. In this work, we propose a Mamba-based architecture for document binarisation, which efficiently handles long sequences by scaling linearly and optimizing memory usage. Additionally, we introduce novel modifications to the skip connections by incorporating Difference of Gaussians (DoG) features, inspired by conventional signal processing techniques. These multiscale high-frequency features enable the model to produce high-quality, detailed outputs.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。