Progress in Camera-Based Document Image Analysis

  • Authors:

  • David Doermann;Jian Liang;Huiping Li

  • Affiliations:

  • -;-;-

  • Venue:
  • ICDAR '03 Proceedings of the Seventh International Conference on Document Analysis and Recognition - Volume 1
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

The increasing availability of high performance, lowpriced, portable digital imaging devices has created atremendous opportunity for supplementing traditionalscanning for document image acquisition. Digitalcameras attached to cellular phones, PDAs, or asstandalone still or video devices are highly mobile andeasy to use; they can capture images of any kind ofdocument including very thick books, historical pages toofragile to touch, and text in scenes; and they are muchmore versatile than desktop scanners. Should robustsolutions to the analysis of documents captured with suchdevices become available, there is clearly a demand frommany domains. Traditional scanner-based documentanalysis techniques provide us with a good reference andstarting point, but they cannot be used directly oncamera-captured images. Camera captured images cansuffer from low resolution, blur, and perspectivedistortion, as well as complex layout and interaction ofthe content and background. In this paper we present asurvey of application domains, technical challenges andsolutions for recognizing documents captured by digitalcameras. We begin by describing typical imaging devicesand the imaging process. We discuss document analysisfrom a single camera-captured image as well as multipleframes and highlight some sample applications underdevelopment and feasible ideas for future development.