Abstract-A new statistical approach based on global typographical features is proposed to the widely neglected problem of font recognition. It aims at the identification of the typeface, weight, slope and size of the text from an image block without any knowledge of the content of that text. The recognition is based on a multivariate Bayesian classifier and operates on a given set of known fonts. The effectiveness of the adopted approach has been experimented on a set of 280 fonts. Font recognition accuracies of about 97 percent were reached on high-quality images. In addition, rates higher than 99.9 percent were obtained for weight and slope detection. Experiments have also shown the system robustness to document language and text content and its sensitivity to text length.

No abstract

No abstract

This paper presents a Convolutional Neural Network (CNN) based page segmentation method for handwritten historical document images. We consider page segmentation as a pixel labeling problem, i.e., each pixel is classified as one of the predefined classes. Traditional methods in this area rely on carefully hand-crafted features or large amounts of prior knowledge. In contrast, we propose to learn features from raw image pixels using a CNN. While many researchers focus on developing deep CNN architectures to solve different problems, we train a simple CNN with only one convolution layer. We show that the simple architecture achieves competitive results against other deep architectures on different public datasets. Experiments also demonstrate the effectiveness and superiority of the proposed method compared to previous methods.

No abstract

No abstract

No abstract

PDF became a very common format for exchanging printable documents. Further, it can be easily generated from the major documents formats, which make a huge number of PDF documents available over the net. However its use is limited to displaying and printing, which considerably reduces the search and retrieval capabilities. For this reason, additional tools have recently appeared that allow to extract the textual content. However their practical use is limited in the sense that the text's reading order is not necessary preserved, especially when handling multi-column documents, or in presence of complex layout. Our thesis is that those tools do not consider the hidden layout and logical structures of documents, which could greatly improve their results.We propose a novel approach to overcome the document content extraction, by merging a) low-level extraction methods applied on PDF files with b) layout analysis performed on a synthetically generated TIFF image. The paper describes the various steps necessary to achieve this task. Finally, we present a first experiment on the restitution of the newspapers' reading order which shows encouraging results.