DocumentCode
2739453
Title
A New Concept of Drive Recording System by Estimating Human Intention Using Particle Filters
Author
Ikoma, Norikazu ; Kawamoto, Kazuhiko ; Kawano, Hiroyuki ; Maeda, Hideaki ; Matsuda, Keisuke ; Akahoshi, K.
Author_Institution
Kyushu Inst. of Technol., Fukuoka
fYear
2007
fDate
5-7 Sept. 2007
Firstpage
409
Lastpage
409
Abstract
A new concept of drive recording system based on estimation of human intention by particle filters has been proposed. It has an omni-directional camera at front ceiling in car cabin to record outside cars and inside driver simultaneously. Outside cars are detected with simple image processing and tracked by finite random set (FRS) state space approach, while driver´s head direction is detected by eigen space method and tracked by elaborated particle filtering technique. Noticed cars noticed of the driver are determined by matching driver´s head direction with the outside cars. The system estimates driver´s prediction on the outside cars´ motion based on the detection result of noticed cars. Hiyari-hatto incident will be detected when large discrepancy between the estimated prediction and the actual situation occurred. Then the system records image sequences around the incident time as well as other signals obtained from sensors of the car.
Keywords
driver information systems; eigenvalues and eigenfunctions; image matching; image sequences; particle filtering (numerical methods); set theory; Hiyari-hatto incident; drive recording system; eigen space method; finite random set; human intention estimation; image processing; image sequences; omnidirectional camera; particle filtering technique; state space approach; Cameras; Filtering; Humans; Image processing; Magnetic heads; Motion detection; Motion estimation; Particle filters; Particle tracking; State-space methods;
fLanguage
English
Publisher
ieee
Conference_Titel
Innovative Computing, Information and Control, 2007. ICICIC '07. Second International Conference on
Conference_Location
Kumamoto
Print_ISBN
0-7695-2882-1
Type
conf
DOI
10.1109/ICICIC.2007.47
Filename
4428051
Link To Document