Presentation 2005-09-16
A home video segmentation method based on regional motion vector classification
Toshikazu KARITSUKA, Satoshi SHIMADA, Masashi MORIMOTO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) A lot of home videos have been taken very easily and therefore stored disorderly these days. It is a forthcoming problem to manage and retrieve those videos easily. This paper proposes a method to segment and label sub-shots of home videos based on the discrimination of camerawork and moving objects. The method uses the assumption that people tend to arrange a focusing object around the center of the video frame when taking a video. Following the assumption, the method divides the video frame into an internal and an external area beforehand, and calculates the feature parameters based on the motion vector of feature points at each area. Each video frame is classified into six categories by the feature parameters using the distance from studied parameters. Experiments found that video frames can be classified more appropriately than the method without frame dividing, and shows the possibility to handle videos with hand swinging and unstable camerawork properly by this method.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Home Video / Video Indexing / Shot Segmentation / Motion Vector
Paper # CQ2005-50,OIS2005-35,IE2005-43
Date of Issue

Conference Information
Committee CQ
Conference Date 2005/9/9(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 Communication Quality (CQ)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A home video segmentation method based on regional motion vector classification
Sub Title (in English)
Keyword(1) Home Video
Keyword(2) Video Indexing
Keyword(3) Shot Segmentation
Keyword(4) Motion Vector
1st Author's Name Toshikazu KARITSUKA
1st Author's Affiliation NTT Cyber Solution Laboratories, NTT Corporation()
2nd Author's Name Satoshi SHIMADA
2nd Author's Affiliation NTT Cyber Solution Laboratories, NTT Corporation
3rd Author's Name Masashi MORIMOTO
3rd Author's Affiliation NTT Cyber Solution Laboratories, NTT Corporation
Date 2005-09-16
Paper # CQ2005-50,OIS2005-35,IE2005-43
Volume (vol) vol.105
Number (no) 282
Page pp.pp.-
#Pages 6
Date of Issue