Deformable template matching within a Bayesian framework for hand-written graphic symbol recognition

Ernest Valveny, Enric Martí

    Research output: Contribution to journalArticleResearchpeer-review

    5 Citations (Scopus)

    Abstract

    We describe a method for hand-drawn symbol recognition based on deformable template matching able to handle uncertainty and imprecision inherent to hand-drawing. Symbols are represented as a set of straight lines and their deformations as geometric transformations of these lines. Matching, however, is done over the original binary image to avoid loss of information during line detection. It is defined as an energy minimization problem, using a Bayesian framework which allows to combine fidelity to ideal shape of the symbol and flexibility to modify the symbol in order to get the best fit to the binary input image. Prior to matching, we find the best global transformation of the symbol to start the recognition process, based on the distance between symbol lines and image lines. We have applied this method to the recognition of dimensions and symbols in architectural floor plans and we show its flexibility to recognize distorted symbols. © Springer-Verlag Berlin Heidelberg 2000.
    Original languageEnglish
    Pages (from-to)193-208
    JournalLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume1941
    Publication statusPublished - 1 Jan 2000

    Fingerprint Dive into the research topics of 'Deformable template matching within a Bayesian framework for hand-written graphic symbol recognition'. Together they form a unique fingerprint.

    Cite this