In the context of historical document analysis, image binarization is a first important step, which separates foreground from background, despite common image degradations, such as faded ink, stains, or bleed-through. Fast binarization has great significance when analyzing vast archives of document images, since even small inefficiencies can quickly accumulate to years of wasted execution time. Therefore, efficient binarization is especially relevant to companies and government institutions, who want to analyze their large collections of document images. The main challenge with this is to speed up the execution performance without affecting the binarization performance. We modify a state-of-the-art binarization algorithm and achieve on average a 3.5 times faster execution performance by correctly mapping this algorithm to a heterogeneous platform, consisting of a CPU and a GPU. Our proposed parameter tuning algorithm additionally improves the execution time for parameter tuning by a factor of 1.7, compared to previous parameter tuning algorithms. We see that for the chosen algorithm, machine learning-based parameter tuning improves the execution performance more than heterogeneous computing, when comparing absolute execution times. © 2018 The Author(s)

In the context of document image analysis, image binarization is an important preprocessing step for other document analysis algorithms, but also relevant on its own by improving the readability of images of historical documents. While historical document image binarization is challenging due to common image degradations, such as bleedthrough, faded ink or stains, achieving good binarization performance in a timely manner is a worthwhile goal to facilitate efficient information extraction from historical documents. In this paper, we propose a recurrent neural network based algorithm using Grid Long Short-Term Memory cells for image binarization, as well as a pseudo F-Measure based weighted loss function. We evaluate the binarization and execution performance of our algorithm for different choices of footprint size, scale factor and loss function. Our experiments show a significant trade-off between binarization time and quality for different footprint sizes. However, we see no statistically significant difference when using different scale factors and only limited differences for different loss functions. Lastly, we compare the binarization performance of our approach with the best performing algorithm in the 2016 handwritten document image binarization contest and show that both algorithms perform equally well.

A major challenge in document image binarization is the large variety in appearance of images from different document collections. This is especially challenging for parameterless, machine learning based binarization algorithms, which require additional ground truth training data to generalize or fine-tune to a new image collection. Reducing this costly labeling effort is relevant to companies and government institutions, which possess many different document image collections. One approach to address this problem is interactive machine learning, which enables a user to guide the fine-tuning process by providing feedback on the produced binarization result.

In this paper, we evaluate the claim that user guided training requires less labeled samples to fine-tune a basic model for binarization to a new image collection. Further, we propose a way to guide user feedback by visualizing the model’s labeling uncertainty and analyze the relationship between model uncertainty and binarization quality. Our experiments show that user feedback biases the model towards favoring foreground labels, which results in less erased text and thus better readability than when training samples are chosen randomly. Additionally, we find that model uncertainty serves as a useful guide for users and explain how the Dunning-Kruger effect prevents model uncertainty from being useful for automated sample selection.