Presentation 2006-09-08
Natural Image Matting with Semi-Supervised Object Extraction
Weiwei DU, Kiichi URAHAMA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In natural image matting where an object is extracted from a photograph with natural background and is composited with another image, a user is required to input strokes specifying both of foreground and background regions for supporting the extraction of the target object. In order to reduce user's inputs, we apply, in this paper, a semi-supervised algorithm for extracting a fuzzy cluster from similarity data to this task, and present a method for extracting an object from a natural image by propagating memberships from seeded pixels to the remaining area according to the similarity between pixel colors transformed with the Fisher's linear discriminant analysis. In our method, rough and coarse strokes are sufficient for users to draw them only either in foreground regions or in background areas. The color of the extracted object is estimated at each pixel based on the membership obtained with this algorithm and the extracted object is composited with another image as a new background.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) natural image matting / semi-supervised object extraction / membership propagation / linear discriminant analysis
Paper # PRMU2006-63
Date of Issue

Conference Information
Committee PRMU
Conference Date 2006/9/1(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 ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Natural Image Matting with Semi-Supervised Object Extraction
Sub Title (in English)
Keyword(1) natural image matting
Keyword(2) semi-supervised object extraction
Keyword(3) membership propagation
Keyword(4) linear discriminant analysis
1st Author's Name Weiwei DU
1st Author's Affiliation Faculty of Design, Kyushu University()
2nd Author's Name Kiichi URAHAMA
2nd Author's Affiliation Faculty of Design, Kyushu University
Date 2006-09-08
Paper # PRMU2006-63
Volume (vol) vol.106
Number (no) 229
Page pp.pp.-
#Pages 6
Date of Issue