Vehicle Appearance Model for Recognition System Considering the Change of Imaging Condition

We have proposed a technique to recognize a vehicle. In this technique, Gaussian Mixture Model (GMM) is adopted as a classifier. Vehicle appearance changed by imaging conditions such as time, weather and so on, and GMM parameters are also changed by imaging conditions. To recognize vehicle accuratel...

Full description

Saved in:
Bibliographic Details
Published in:Journal of advanced computational intelligence and intelligent informatics 2009-07, Vol.13 (4), p.463-469
Main Authors: Kuwabara, Keiji, Yano, Yoshikazu, Okuma, Shigeru
Format: Article
Language:English
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We have proposed a technique to recognize a vehicle. In this technique, Gaussian Mixture Model (GMM) is adopted as a classifier. Vehicle appearance changed by imaging conditions such as time, weather and so on, and GMM parameters are also changed by imaging conditions. To recognize vehicle accurately, we have prepared some GMM tuned with the imaging conditions. On the other hand, it is impossible to prepare GMM because imaging condition changes successively. In this paper, we propose a method for estimating GMM and for training GMM parameters which reflect the successive change of imaging condition. Experimental results show that GMM parameters are estimated accurately and training of GMM are speeded up by proposed method.
ISSN:1343-0130
1883-8014
DOI:10.20965/jaciii.2009.p0463