Open AccessModified chain‐code‐based object recognition
Author(s) -
Lee Daeha,
Kim SoonJa
Publication year - 2015
Publication title -
electronics letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.375
H-Index - 146
ISSN - 1350-911X
DOI - 10.1049/el.2015.1019
Subject(s) - chain code , encoding (memory) , computer science , encode , code (set theory) , object (grammar) , chain (unit) , feature (linguistics) , artificial intelligence , compression (physics) , data compression , computer vision , pattern recognition (psychology) , object oriented programming , image (mathematics) , programming language , biochemistry , chemistry , physics , linguistics , philosophy , materials science , set (abstract data type) , composite material , astronomy , gene
Chain code, a data compression technique, has been widely used to encode the contour information of an object or gesture. Although this chain code is effective for data compression, it is impossible to avoid the duplication of borders. That is, even though they represent different objects, the encoding results can be made similar. The vector chain code (VCC), which contains the direction and distance information to solve the problems of the existing chain code is proposed. Also, the integrated VCC is composed of internal and external class‐based VCCs to support the encoding of objects that have multiple feature points. Experiments demonstrate that the proposed VCC achieves an improved recognition rate in comparison to that achieved by the existing chain code.