Multi-oriented Handwritten Annotations Extraction from Scanned Documents

Benjlaiel, Mohamed; Mullot, Rémy; Alimi, Adel M.

doi:10.1109/das.2014.17

Cited by 7 publications

(3 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Geometrical features such as distance, area, and density are also commonly used to extract text-lines [23,70] Classification These algorithms classify structural elements (pixels, letters, text-lines...) from a set of learned features. Some of them have been successively applied to separate handwritten annotations from printed text by using connected component and patch level features [53,54], shape context features [22] or more traditional features [8]. In [36], structure detection of degraded newspaper archives is achieved by localizing titles, textlines, background, separators and noise using a Conditional Random Field.…”

Section: Bottom-up or Data-driven Strategiesmentioning

confidence: 99%

Combination of deep neural networks and logical rules for record segmentation in historical handwritten registers using few examples

Tarride

Lemaitre

Coüasnon

et al. 2021

IJDAR

View full text Add to dashboard Cite

This work focuses on the layout analysis of historical handwritten registers, in which local religious ceremonies were recorded. The aim of this work is to delimit each record in these registers. To this end, two approaches are proposed. Firstly, object detection networks are explored, as three state-of-the-art architectures are compared. Further experiments are then conducted on Mask R-CNN, as it yields the best performance. Secondly, we introduce and investigate Deep Syntax, a hybrid system that takes advantages of recurrent patterns to delimit each record, by combining ushaped networks and logical rules. Finally, these two approaches are evaluated on 3708 French records (16-18th centuries), as well as on the Esposalles public database, containing 253 Spanish records (17th century). While both systems perform well on homogeneous documents, we observe a significant drop in performance with Mask R-CNN on heterogeneous documents, especially when trained on a non-representative subset. By contrast, Deep Syntax relies on steady patterns, and is therefore able to process a wider range of documents with less training data. Not only Deep Syntax produces 15% more match configurations and reduces the ZoneMap surface error metric by 30% when both systems are trained on 120 images, but it also outperforms Mask R-CNN when trained on a database three times smaller. As Deep Syntax generalizes better, we believe it can be used in the context of massive document processing, as collecting and annotating a sufficiently large and representative set of training data is not always achievable.

show abstract

Section: Bottom-up or Data-driven Strategiesmentioning

confidence: 99%

Combination of deep neural networks and logical rules for record segmentation in historical handwritten registers using few examples

Tarride

Lemaitre

Coüasnon

et al. 2021

IJDAR

View full text Add to dashboard Cite

show abstract

“…Black and white algorithms. Among them, use only binary images [9,29,[90][91][92][93][94][95][96][97][98][99].…”

Section: Feature Classificationmentioning

confidence: 99%

A comprehensive survey of mostly textual document segmentation algorithms since 2008

Eskenazi

Gomez‐Krämer

Ogier

2017

Pattern Recognition

View full text Add to dashboard Cite

International audienceIn document image analysis, segmentation is the task that identifies the regions of a document. The increasing number of applications of document analysis requires a good knowledge of the available technologies. This survey highlights the variety of the approaches that have been proposed for document image segmentation since 2008. It provides a clear typology of documents and of document image segmentation algorithms. We also discuss the technical limitations of these algorithms, the way they are evaluated and the general trends of the community

show abstract

“…A Fisher classifier is then employed to associate the pseudo-word to the handwritten or the printed script. Fourier descriptors, Gabor filters, and Hu moments was extracted in [21]. A separate k-NN classifier was trained for each descriptor features, k-NN outputs were then combined using a simple majority vote.…”

Section: B Classificationmentioning

confidence: 99%

Neighborhood label extension for handwritten/printed text separation in Arabic documents

Awal¹,

Belaïd²

2017

2017 1st International Workshop on Arabic Script Analysis and Recognition (ASAR)

View full text Add to dashboard Cite

This paper addresses the problem of handwritten and printed text separation in Arabic document images. The objective is to extract handwritten text from other parts of the document. This allows the application, in a second time, of a specialized processing on the extracted handwritten part or even on the printed one. Documents are first preprocessed in order to remove eventual noise and correct document orientation. Then, the document is segmented into pseudo-lines that are segmented in turn into pseudo-words. A local classification step, using a Gaussian kernel SVM, associates each pseudo-word into handwritten or printed classes. This label is then propagated in the pseudo-word's neighborhood in order to recover from classification errors. The proposed methodology has been tested on a set of public real Arabic documents achieving a separation rate of around 90%.

show abstract

Multi-oriented Handwritten Annotations Extraction from Scanned Documents

Cited by 7 publications

References 15 publications

Combination of deep neural networks and logical rules for record segmentation in historical handwritten registers using few examples

Combination of deep neural networks and logical rules for record segmentation in historical handwritten registers using few examples

A comprehensive survey of mostly textual document segmentation algorithms since 2008

Neighborhood label extension for handwritten/printed text separation in Arabic documents

Contact Info

Product

Resources

About