Sprungmarken

Service navigation

Hauptnavigation

You are here:

Bereichsnavigation

Hauptinhalt

Seminar Document Analysis and Recognition - SS 2011


Kick-off meeting (including mandatory registration):
April 6, 2011, 2 PM, OH-8, R.108 (IRF Seminar Room)

Introduction

One of the largest application areas of pattern recognition methods is the field of document analysis and recognition. The problems addressed range from optical character recognition (i.e. reading machine printed texts), postal address reading, handwriting recognition, processing of historical documents, and document image retrieval to pen computing for man-machine interaction. The techniques applied for solving these problems usually consist of a central method known from pattern recognition augmented by several application-specific pre- or post-processing techniques. In this seminar we want to explore the field of document analysis and recognition by having a closer look at the different tasks investigated which become even more challenging when considering the large variety of document types and scripts that can or need to be taken into account.

Mandatory Module (Pflichtmodul INF-MSc-102: Seminar) for Master (Applied) Computer Science

Topical focus areas (Schwerpunktgebiete): 2 (..., Embedded Systems, ...), 7 (Intelligent Systems)

Tentative list of topics:

A) General foundations

  • The worlds writing systems
  • Cognitive processes in human reading
B) Machine reading of text
  • Text detection in images and video
  • Line extraction from text documents
  • Preprocessing and normalization
  • Recognition of machine printed text
  • Handwriting recognition (HMM-based, offline)
  • Recognition of "non-standard" scripts, e.g., Chinese, Arabic, Indian (Bangla, Devnagari, ...)
  • Recognition of special text categories, e.g., digits, bank cheques, license plates
C) Document analysis
  • Document layout analysis and document classification
  • Postal document analysis / recognition
  • Script identification
  • Writer identification
  • Camera-based document analysis
  • Pen computing (i.e. online recognition of handwritten text, mathematical formulas, drawings, ...)
  • Analysis of historical documents
  • Document image retrieval
Organizers: Gernot A. Fink and Szilard Vajda