Document Image Binarization with Fully Convolutional Neural Networks

TL;DR

This paper introduces a novel fully convolutional neural network architecture for binarizing degraded historical manuscript images, achieving state-of-the-art results across multiple datasets and adaptable to different document types.

Contribution

The paper proposes a multi-scale FCN architecture trained with a continuous Pseudo F-measure, outperforming previous methods on several benchmarks and demonstrating versatility across document domains.

Findings

Ensemble of FCNs outperforms previous winners on DIBCO datasets.

The method generalizes well to Palm Leaf Manuscripts.

Performance depends on hyperparameters, training data size, and input features.

Abstract

Binarization of degraded historical manuscript images is an important pre-processing step for many document processing tasks. We formulate binarization as a pixel classification learning task and apply a novel Fully Convolutional Network (FCN) architecture that operates at multiple image scales, including full resolution. The FCN is trained to optimize a continuous version of the Pseudo F-measure metric and an ensemble of FCNs outperform the competition winners on 4 of 7 DIBCO competitions. This same binarization technique can also be applied to different domains such as Palm Leaf Manuscripts with good performance. We analyze the performance of the proposed model w.r.t. the architectural hyperparameters, size and diversity of training data, and the input features chosen.