Fully Convolutional Neural Networks for Page Segmentation of Historical Document Images

TL;DR

This paper introduces a fully convolutional neural network for efficient, single-step segmentation of historical document images, outperforming existing methods and utilizing a novel, ambiguity-independent evaluation metric.

Contribution

The paper presents a novel FCN model for document segmentation that learns directly from raw pixels and introduces the Foreground Pixel Accuracy metric for unbiased evaluation.

Findings

The FCN outperforms existing segmentation methods on public datasets.

The proposed FgPA metric provides a more reliable evaluation independent of ambiguous ground truth.

The model processes a single page in one step, improving speed and simplicity.

Abstract

We propose a high-performance fully convolutional neural network (FCN) for historical document segmentation that is designed to process a single page in one step. The advantage of this model beside its speed is its ability to directly learn from raw pixels instead of using preprocessing steps e. g. feature computation or superpixel generation. We show that this network yields better results than existing methods on different public data sets. For evaluation of this model we introduce a novel metric that is independent of ambiguous ground truth called Foreground Pixel Accuracy (FgPA). This pixel based measure only counts foreground pixels in the binarized page, any background pixel is omitted. The major advantage of this metric is, that it enables researchers to compare different segmentation methods on their ability to successfully segment text or pictures and not on their ability to learn and possibly overfit the peculiarities of an ambiguous hand-made ground truth segmentation.