Presentation 2015-02-23
Image Feature Descriptor of Rotation-Invariant Gradient Orientation for Pedestrian Detection under Changes of Posture
Yuta HORIKAWA, Kousuke MATSUSHIMA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Pedestrian recognition is an important subject of research in automotive safe driving support technology. Combining image feature descriptors and classifiers is a classical method for pedestrian recognition. HOG are one of image feature descriptors and had shown high performance in pedestrian recognition. Various feature descriptors improving HOG have been proposing to date. The more similar feature vectors extracted from pedestrian learning images feature vectors extracted from ROI images are get, the more ROI images are recognized as pedestrian. However, it sometimes occurs that feature vectors extracted from pedestrian images aren't similar to feature vectors extracted form pedestrian learning images, because person can take various poses. One of the poses is caused by rotating person posture. This paper proposes rotation-invariant oriented gradient feature descriptors for rotating changes of posture.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) image feature descriptor / pedestrian detection / pattern recognition
Paper # ITS2014-42,IE2014-69
Date of Issue

Conference Information
Committee ITS
Conference Date 2015/2/16(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Intelligent Transport Systems Technology (ITS)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Image Feature Descriptor of Rotation-Invariant Gradient Orientation for Pedestrian Detection under Changes of Posture
Sub Title (in English)
Keyword(1) image feature descriptor
Keyword(2) pedestrian detection
Keyword(3) pattern recognition
1st Author's Name Yuta HORIKAWA
1st Author's Affiliation Mechanical and Electrical Systems Engineering Advanced Course, Advanced Engineering School, Kurume National College of Technorogy()
2nd Author's Name Kousuke MATSUSHIMA
2nd Author's Affiliation Department of Control and Information Systems Engineering, Kurume National College of Technorogy
Date 2015-02-23
Paper # ITS2014-42,IE2014-69
Volume (vol) vol.114
Number (no) 459
Page pp.pp.-
#Pages 6
Date of Issue