A visual attention and tracking design for a robotic vision system

Chia How Lin*, Yen Ting Chen, Kai-Tai Song

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Scopus citations

Abstract

Many robotic vision systems suffer from unsatisfied recognition rate and high computational load. A solution to the design of a fast and sufficiently accurate robotic vision system requires urgent attention. This paper presents a novel image processing architecture for visual attention and tracking applications using a series of low-level image processing units. The proposed system features a human-mimic way to segment and keep tracking of interested objects. These simplification and tracking schemes help to enhance both the speed and correctness of high level vision applications such as object detection and recognition. Furthermore, the proposed system can be executed under low frame rates, and can adapt itself to changing environments, which often cause problems in practical robotic applications. The experimental results show that the system is robust against pose translation, camera motion, motion blur, and temporal occlusion.

Original languageEnglish
Title of host publication2009 IEEE Workshop on Advanced Robotics and Its Social Impacts, ARSO2009 - Workshop Proceedings
Pages30-35
Number of pages6
DOIs
StatePublished - 1 Dec 2009
Event2009 IEEE Workshop on Advanced Robotics and Its Social Impacts, ARSO2009 - Tokyo, Japan
Duration: 23 Nov 200925 Nov 2009

Publication series

NameProceedings of IEEE Workshop on Advanced Robotics and its Social Impacts, ARSO
ISSN (Print)2162-7568
ISSN (Electronic)2162-7576

Conference

Conference2009 IEEE Workshop on Advanced Robotics and Its Social Impacts, ARSO2009
CountryJapan
CityTokyo
Period23/11/0925/11/09

Fingerprint Dive into the research topics of 'A visual attention and tracking design for a robotic vision system'. Together they form a unique fingerprint.

Cite this