Entropy Quantifiers Useful for Establishing Equivalence between Text Document Images

There are many requirements in document image analysis, which warrant understanding the equivalence of document images if possible without OCRing the text contents and in some cases OCRs do not exist. In this paper we propose to employ the entropy notion to' feel' the text content in a doc...

Full description

Saved in:
Bibliographic Details
Main Authors: Gowda, S.D., Nagabhushan, P.
Format: Conference Proceeding
Language:English
Subjects:
Energy measurement
Entropy
Image analysis
Image coding
Image generation
Image recognition
Optical character recognition software
Pixel
Position measurement
Shape
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:There are many requirements in document image analysis, which warrant understanding the equivalence of document images if possible without OCRing the text contents and in some cases OCRs do not exist. In this paper we propose to employ the entropy notion to' feel' the text content in a document image without actually reading it, and hence establish the equivalence or otherwise of two corresponding text components (line/word/character). We introduce Conventional Entropy Quantifier (CEQ) and also define Modified Entropy Quantifier (MEQ) to measure the energy content in the components. The results of experiments performed at line, word and character level are reported. These initial steps in the sequel are expected to establish the equivalence between the two text document images.
DOI:10.1109/ICCIMA.2007.304