Ameur Bensefia

A writer identification and verification system

In this paper, we show that both the writer identification and the writer verification tasks can be carried out using local features such as graphemes extracted from the segmentation of cursive handwriting. We thus enlarge the scope of the possible use of these two tasks which have been, up to now, mainly evaluated on script handwritings. A textual based Information Retrieval model is used for the writer identification stage. This allows the use of a particular feature space based on feature frequencies. Image queries are handwritten documents projected in this feature space. The approach achieves 95% correct identification on the PSI_DataBase and 86% on the IAM_DataBase. Then writer hypothesis retrieved are analysed during a verification phase. We call upon a mutual information criterion to verify that two documents may have been produced by the same writer or not. Hypothesis testing is used for this purpose. The proposed method is first scaled on the PSI_DataBase then evaluated on the IAM_DataBase. On both databases, similar performance of nearly 96% correct verification is reported, thus making the approach general and very promising for large scale applications in the domain of handwritten document querying and writer verification.

Information retrieval based writer identification

This communication deals with the Writer Identificationtask. Our previous work has shown the interest of usingthe graphemes as features for describing the individualproperties of Handwriting. We propose here to exploit thesame feature set but using an information retrievalparadigm to describe and compare the handwritten queryto each sample of handwriting in the database. Using thistechnique the image processing stage is performed onlyonce and before the retrieval process can take place, thusleading to a significant saving in the computation of eachquery response, compared to our initial proposition. Themethod has been tested on two handwritten databases.The first one has been collected from 88 different writersat PSI Lab. while the second one contains 39 writers fromthe original correspondence of Emile Zola, a famousFrench novelist of the last 19th century. We also analyzethe proposed method when using concatenation ofgraphemes (bi and tri-gramme) as features.

Writer identification by writer's invariants

This communication deals with the problem of writer identification. If the assumption of writing individuality is true then graphical fragments that constitute it should be individual too. Therefore we propose a morphological grapheme based analysis to make writer identification. Template Matching is the core of the approach. The redundancy of the individual patterns in a writing, defined as the writers invariants, allows to compress the handwritten texts while maintaining good identification performance. Two series of tests are reported. The first series is designed to evaluate the relevance of our approach of identification on a basis of 88 writers by evaluating the influence of the text representation (with or without invariants) on the quality of the method. The method gives about 97,7% of correct identification when using large compressed samples of handwriting. The second series of tests is designed to evaluate the influence of the sample size of the writing to be identified on the quality of the method. It is shown that writer identification can reach a correct identification rate of 92,9% using only samples of 50 graphemes of each writing.

Handwriting analysis for writer verification

This communication deals with the writer verification task. This task consists in deciding whether two handwritten samples have been written by the same writer or not. Handwritings are first characterized by the graphemes that have been segmented by a segmentation procedure. Handwritten samples are then analysed according to two different procedures. Text samples are described in a feature space common to the two writers. The statistic of a mutual information criteria allows to build a robust hypothesis test. In the case of small samples of handwritings such as single words, the Levenstein distance is used to build a second hypothesis test. The two approaches are evaluated on a PSI database as well as the IAM database.

Handwritten text recognition through writer adaptation

Handwritten text recognition is a problem rarely studied out of specific applications for which lexical knowledge can constrain the vocabulary to a limited one. In the case of handwritten text recognition, additional information can be exploited to characterize the specificity of the writing. This knowledge can help the recognition system to find coherent solutions from both the lexical and the morphological points of view. We present the principles of a handwritten text recognition system based on the online learning of the writer shapes. The proposed scheme is shown to improve the recognition rates on a sample of fifteen writings, unknown to the system.

Documents manuscrits et recherche d'information

Nous presentons un modele de recherche d’information visuelle adapte a la navigation et l’interrogation de bases de documents manuscrits numerises. Nous considerons ces documents du point de vue de leur contenu graphique, ce qui inscrit cette problematique dans un cadre d’identification du scripteur. Un certain nombre de travaux ont aborde ce probleme d’identification du scripteur, le plus souvent en s’appuyant sur des techniques d’analyse de textures pour caracteriser les ecritures. L’originalite des travaux que nous presentons tient au fait que nous fondons notre demarche sur une technique de recherche d’information en utilisant une description specifique a l’ecriture manuscrite. L’approche est evaluee sur deux bases de documents manuscrits: une base creee au laboratoire et une base du patrimoine litteraire constituee des correspondances de Zola.