Guest Editorial

Day in, day out, billions of business and finance documents have to be processed in order to make time-critical information available for corporate workflow. During recent years document analysis and recognition technology have been introduced to effectively support information capturing and archiving for such items as bank checks, credit card payment slips, tax documents, invoices, forms, delivery notes, maps and graphics. The aim of this special issue is to look at document analysis from the application perspective and to give insight into where document analysis stands, what the current application-oriented research topics are, and where future research should go. Only a fraction of the papers submitted for this special issue could be selected for publication. The reviewing procedure was done by up to four experts per paper, partly in a two-step process: Revised papers were again reviewed by the experts in order to guarantee the highest possible quality and understandability. Ten of the chosen papers excel in their especially high relevance to application. They will be published in two volumes; the present one is the first. More papers will be published in a regular issue of the IJDAR journal. The first two papers address one of the first largescale application fields of documents analysis: the automatic processing of bank checks. The paper, “Industrial bank check processing: the A2iA CheckReader TM” by N. Gorski, V. Anisimov, E. Augustin, O. Baret, and S. Maximov, describes a commercial bank check recognition system. This system is designed to replace human operators in payment document clearing and is running successfully in many locations. The paper provides a detailed description of the architecture and features of this system and pays special attention to the country concerned and language-specific problems. It also gives an overview of the implemented document processing techniques and approaches. While the A2iA CheckReader focuses on European languages, the paper, “Segmentation and recognition of Chinese bank cheque amounts” by M.L. Yu, P.C.K. Kwok, C.H. Leung, and K.W. Tse, proposes a Chinese bank check recognition system. The subsystems, consisting of preprocessing, segmentation and recognition, are described in detail. Special attention is paid to a newly developed grammar checker for Chinese bank check amounts. A supportive system used for forensic studies of handwritten documents by the German federal police department (the “Bundeskriminalamt”) is described in the paper “A computer-based system to support forensic studies on handwritten documents” by K. Franke and M. Köppen. The goal of the system is to extract the original handwriting from various backgrounds to the extent that this is possible. This task contrasts to that of bank check readers, where the background is relatively stable. In order to meet this challenge the paper proposes an open layered framework that covers adaptive abilities on the parametric, operational and algorithmic level. An embedded module using genetic programming is proposed for the ad-hoc generation of specific background filters. Another important application field of document analysis is the classification of documents as described in the paper “Classification of document pages using structurebased features” by C. Shin, D. Doermann, and A. Rosenfeld. The paper describes an approach for classifying documents according to type and genre based on their visual appearance. Documents in the approach are represented by various layout-based features which can be computed without any class knowledge. Classification takes a supervised approach, using tree classifiers and self-organizing maps. The paper pays special attention to the subjectivity of relevance judgments made by humans in the training of classifiers. A knowledge-based system for the automated processing of business letters is proposed in the last paper, “Leveraging corporate context within knowledge-based document analysis and understanding” by C. Wenzel and H. Maus. The proposed system “VOPR” integrates a commercial workflow system in document processing. Incoming documents are analyzed according to the context available in the workflow management system. Documents and results are assigned to the appropriate workflow resource. As mentioned above, this is the first of two volumes of the special issue on Document Analysis for Office Systems. Further interesting papers can be found in the second volume, as well as in one of the next regular issues of the IJDAR journal. We would already like to take this opportunity to thank all authors for the many papers submitted, of which we could only publish a fraction. Our special thanks goes to all reviewers, who spent so much time helping to make these papers as useful and understandable as possible.