By Topic

Document Image Processing and Multimedia Environments, IEE Colloquium on

Date 2 Nov 1995

Filter Results

Displaying Results 1 - 16 of 16
  • Segmentation and classification of document images

    Publication Year: 1995 , Page(s): 16/1 - 16/7
    Cited by:  Patents (3)
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (516 KB)  

    There is a significant and growing need to convert documents from printed paper to an electronic form. Document image analysis is concerned with the segmentation of the document image into regions of interest, their description, and the classification of the regions according to the type of their contents. A new unified approach to page segmentation and classification, based on the description of the background with tiles, is presented. The segmentation method is flexible to successfully analyse and describe regions in complicated layouts where other methods fail. Images with severe skew are handled equally well with no additional computations. The classification is based on textural features which are derived by simple calculations from the representation of space in the regions, produced during the segmentation process. This is a considerable advantage over previous methods where extra image accesses and lengthy computations are necessary. Overall, the whole approach of segmentation and classification by white tiles is fast and efficient as no time-consuming processes are required View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Low- and high-level approaches to optical music score recognition

    Publication Year: 1995 , Page(s): 3/1 - 3/6
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (456 KB)  

    The computer has become an increasingly important device in music. It can not only generate sound but is also able to perform time consuming and repetitive tasks, such as transposition and part extraction, with speed and accuracy. However, a score must be represented in a machine readable format before any operation can be carried out. Current input methods, such as using an electronic keyboard, are time consuming and require human intervention. Optical music recognition (OMR) provides an interesting, efficient and automatic method to transform paper-based music scores into a machine representation. The authors outline the techniques for pre-processing and discuss the heuristic and musical rules employed to enhance recognition. A spin-off application that makes use of the intermediate results to enhance stave lines is also presented. The authors concentrate on the techniques used for time-signature detection, discuss the application of frequently-found rhythmical patterns to clarify the results of OMR, and propose possible enhancements using such knowledge. They believe that domain-knowledge enhancement is essential for complex document analysis and recognition. Other possible areas of development include melodic, harmonic and stylistic analysis to improve recognition results further View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Distributed multimedia information systems-imaging and networking issues

    Publication Year: 1995 , Page(s): 9/1 - 9/6
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (392 KB)  

    Distributed multimedia information systems (DMIS) are the emerging technology which provide its users with the facility to exchange compound documents comprising of text, images, graphics and sound. They mainly consist of clusters of workstations linked via networks. These systems cater for store-and-forward and/or real-time information. Recent developments in distributed computing and applications oriented toward image processing require high speed networks to transfer large amounts of data with minimum delay. This paper reviews some high speed networks and some image compression strategies for coping with the large amount of data View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • On computerisation and representation of characters

    Publication Year: 1995 , Page(s): 11/1 - 11/9
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (468 KB)  

    The commercial need to computerise contours of objects has intensified over recent years as more and more applications endeavour to seek the benefits of visualisation. The primary aim has been to gain a computerised model which closely resembles the original shape and form. The digital form lends itself to be processed by fast processors, and their output to be displayed on graphic devices such as screen displays and laser printers. The benefits of the digital form are also extended to cater for numerically controlled machines. These are employed to produce outlines through cutting or engraving on paper, metal, plastic or wood. When it comes to digitising the outlines of characters, additional considerations have to be included in the modelling, design, phase. The aim of these is to capture the uniqueness and the distinct features of a font, which traditionally were realised by a proficient artisan's hand and eye. Modern approaches endeavour to gain the same impact by using a set of pre-defined specifications in the design process. The goal of which is not just to model accurately a desired font, but also to ensure that the displayed version is both legible and aesthetically acceptable. This paper provides an insight about the design methods used to both model and represent the contours of Latin characters and fonts. It describes the role of the spline as a means of realising the modelling aspect. To this regard, the spline descriptions used by Postscript (Bezier cubic) and TrueType (parabolic) are looked at, and a comparative analysis made. The discussion throughout is levelled at providing a practical understanding of the subject matter View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A computationally efficient technique for discriminating between hand-written and printed text

    Publication Year: 1995 , Page(s): 17/1 - 17/7
    Cited by:  Papers (3)  |  Patents (3)
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (264 KB)  

    During the processing of documents from a range of sources, it is useful to be able to discriminate documents with hand-written text from those with printed text. This allows the two to be processed in different ways; for example the typed text using a high speed automated OCR reader and the hand-written text manually, so optimising the use of the expensive OCR reading machine. The paper describes a computationally efficient technique for discriminating between hand-written and printed text on mail. The method employs two stages of processing. The first involves the extraction of a number of low-level features from an image of the sample text, while the second stage is a parametric classification operation, which employs a relatively simple feedforward multilayer perception neural network. The processing involved in each of these stages is described in detail. In addition, results which were obtained when using the optimised techniques to discriminate between hand-written and printed addresses are presented View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Image processing for compound documents

    Publication Year: 1995 , Page(s): 1/1 - 1/8
    Cited by:  Patents (2)
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (480 KB)  

    Until recently, a lot of document image processing (DIP) research has concentrated on the recognition of characters and shapes in document. Whilst the problems of text only based documents have largely been solved, there is a greater need to address the image processing problems posed by mixed-media or multimedia documents. Such documents, commonly referred to as compound documents containing a mixture of text, graphics, line-art, sound annotation, images etc. produce the problem of distinguishing the type of image processing required and in what sector of the document. This paper discusses the processes involved in transforming paper documents into electronic format and gives an example of an application involving the transformation of circuit diagrams to electronic CAD (ECAD) format View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Content based navigation in an open hypermedia environment

    Publication Year: 1995 , Page(s): 5/1 - 5/6
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (640 KB)  

    The Microcosm architecture has been extended to provide a framework for content based navigation and retrieval from non-text media. Exemplar modules for use with images, video and sound have been implemented to provide some basic general purpose CBN and CBR facilities. Examples of experimental hypermedia systems which the authors are developing using the Microcosm/MAVIS software include a product catalogue for a large retailer which contains product images and text, and an archaeological information system containing large numbers of artefact images and associated documentation. Current modules use statistical pattern matching techniques for signature matching to provide general purpose navigational tools. Future work is expected include the development of faster indexing strategies and more intelligent navigation aids including the development signature modules which use of prior knowledge of expected objects to provide semantically based, applications specific navigation View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • On-line legal aid: Markov chain model for efficient retrieval of legal documents

    Publication Year: 1995 , Page(s): 15/1 - 15/7
    Cited by:  Patents (3)
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (632 KB)  

    It is widely accepted that, with large databases, the key to good performance is effective data-clustering. In any large document database clustering is essential for efficient search, browse and therefore retrieval. Cluster analysis allows the identification of groups, or clusters, of similar objects in multi-dimensional space. Conventional document retrieval systems involve the matching of a query against individual documents, whereas a clustered search compares a query with clusters of documents, thereby achieving efficient retrieval. In most document databases periodic updating of clusters is required due to the dynamic nature of a database. Experimental evidence, however shows that clustered searches are substantially less effective than conventional searches of corresponding non-clustered documents. We investigate the present clustering criteria and its drawbacks. We propose a new approach to clustering and justify the reasons why this new approach should be tested and (if proved beneficial) adopted View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • UCAS and electronic delivery of student application forms

    Publication Year: 1995 , Page(s): 7/1 - 7/8
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (480 KB)  

    The aim of the Universities and Colleges Admissions Service (UCAS) is to fill every available higher education course in the UK with the most appropriate students as quickly and cost-effectively as possible. With increasing numbers of applicants, realising this vision will mean delivering student application forms electronically to all the institutions it serves, which involves moving from a paper-based to an electronic form. Such change cannot happen overnight, and needs careful planning. To achieve this, three separate projects are underway, each of which will play a part in turning the vision into reality. These are the ATAPS, Electronic Application and ICR projects. The purpose of this paper is to give an insight into the document management technology which will be used View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Rapid searching of massive dictionaries given uncertain inputs

    Publication Year: 1995 , Page(s): 12/1 - 12/8
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (488 KB)  

    A new method of searching large dictionaries given uncertain inputs is described, based on the lazy evaluation of a syntactic neural network (SNN). The new method is shown to significantly outperform a conventional tree-based method for large dictionaries (e.g. in excess of 100000 entries). Results are presented for the problem of recognising UK postcodes using dictionary sizes of up to 1 million entries. Most significantly, it is demonstrated that the SNN actually gets faster as more data is loaded into it View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Interaction with documents using video scanning and projection

    Publication Year: 1995 , Page(s): 18/1 - 18/6
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (456 KB)  

    The paper describes recent work towards a desk-top environment for the interactive manipulation of paper and electronic forms of documents, with an emphasis placed on its physical realisation. We present the methods used for calibration of the camera-projector system and outline a few applications that are made possible with the use of this particular multimedia environment View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Text block recognition from TIFF images

    Publication Year: 1995 , Page(s): 4/1 - 4/6
    Cited by:  Papers (1)  |  Patents (3)
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (584 KB)  

    The reproduction of a scanned document should include not only the optical character recognition of text, but also the structure of that text on the page and the appearance of that text itself (i.e. font recognition). This is paper presents an algorithm which structurally recognises the text of a page image. The method is based upon the “Docstrum plot” algorithm by L.O'Gorman (1993). Modifications have been made to O'Gorman's algorithm which render very good results at identifying paragraphs and lines in particular. The algorithm implementation can, to a limited degree, describe the logical relationship of the text elements of the original page. The limitations of the algorithm are due to the lack of information available without OCR and font technology incorporated into the algorithm implementation. The algorithm implementation has a graphical interface which portrays the state of the algorithm during the process of decomposition View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Pseudo 2-dimensional hidden Markov models for document restoration

    Publication Year: 1995 , Page(s): 10/1 - 10/6
    Cited by:  Patents (2)
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (388 KB)  

    Ancient documents suffer from many types of degradation, ranging from dirt and staining to damage and fading. Also archaeological and historical texts are hand written. This article explains how hidden Markov models may be used in the image processing of these documents View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The credit card as a mass storage medium

    Publication Year: 1995 , Page(s): 13/1 - 13/5
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (272 KB)  

    Current credit cards store about 140 bytes of information on a magnetic stripe over three tracks. A system has been devised to store and retrieve more than 2 kbytes on a card while maintaining compatibility with current cards. To achieve this a combination of a multi-track read head and new soft decision techniques have been used View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A configurable toolkit approach to handwritten forms recognition

    Publication Year: 1995 , Page(s): 2/1 - 2/6
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (812 KB)  

    Describes a software architecture for configuring and integrating a toolkit of standard OCR components with user-specified form-specific contextual knowledge. The authors illustrate the use of the toolkit by describing two different handwritten form applications, a DSS-P14 tax form and an insurance proposal form. They show the architecture of the overall system, the components required to build it, and present initial results obtained after evaluating the performance of the system using representative forms databases View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Self service document processing for banking automation

    Publication Year: 1995 , Page(s): 8/1 - 8/5
    Cited by:  Patents (3)
    Save to Project icon | Click to expandQuick Abstract | PDF file iconPDF (316 KB)  

    The high street cash dispenser or automated teller machine (ATM) was one of the major technological successes of the eighties. A recent major advance in the development of the ATM has been to extend the range of services available to the consumer to include payment transactions eg. bill payment and cheque deposit. These additional services now make it possible for as ATM to provide the consumer with a complete suite of automated financial transactions at convenient locations 24 hours per day. The result of AT&Ts investment and development effort in this field has been the creation of a new self service document processing module (DPM) that enables ATMs to accept and process a wide range of documents including bills, cheques and giros from almost any country in the world. Considerable technical challenges were faced by AT&T during the development ofthe DPM. Central to this solution is the document processing transaction that requires the payment ATM to accept a document from the consumer and read the relevant information from it so the data can be electronically transacted. This paper describes some of the technical issues and associated solutions that have been developed by AT&T to automate document based payment transactions in a self service terminal View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.