Presentation 1996/5/17
Split-and-Merge Contour Models Based on Crossing Detection for Tracking Multiple Moving Objects
Shoichi ARAKI, Naokazu YOKOYA, Hidehiko IWASA, Haruo TAKEMURA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Active contour models (snakes) are useful for tracking contours of deformable objects compared with conventional approaches based on segmentation and motion measurement. However, Snakes can not follow correct contours of multiple objects when they overlap. This report describes a new split-and-merge contour model based on detecting self-crossing and mutual-crossing of the models. The proposed method can find occlusions of multiple objects by detecting mutual-crossing of different models which extract different objects. Multiple different models which are mutually crossing are merged into a single model. The merged model is split into multiple models when overlapping objects leave away from each other. In order to prove the effectiveness of the method, we apply our model to extracting walking persons in a room scene.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) active contour model / split-and-merge / type model / moving object tracking / optical flow
Paper # PRMU96-21
Date of Issue

Conference Information
Committee PRMU
Conference Date 1996/5/17(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 Pattern Recognition and Media Understanding (PRMU)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Split-and-Merge Contour Models Based on Crossing Detection for Tracking Multiple Moving Objects
Sub Title (in English)
Keyword(1) active contour model
Keyword(2) split-and-merge
Keyword(3) type model
Keyword(4) moving object tracking
Keyword(5) optical flow
1st Author's Name Shoichi ARAKI
1st Author's Affiliation Central Research Laboratories, Matsushita Electric Industrial Co., LTD.()
2nd Author's Name Naokazu YOKOYA
2nd Author's Affiliation Graduate School of Information Science, Nara Institute of Science and Technology
3rd Author's Name Hidehiko IWASA
3rd Author's Affiliation Graduate School of Information Science, Nara Institute of Science and Technology
4th Author's Name Haruo TAKEMURA
4th Author's Affiliation Graduate School of Information Science, Nara Institute of Science and Technology
Date 1996/5/17
Paper # PRMU96-21
Volume (vol) vol.96
Number (no) 41
Page pp.pp.-
#Pages 8
Date of Issue