Presentation | 2006-09-09 Object Recognition using Relation Models Daisuke KATSUMA, Daisuke SHIMIZU, Mutsumi WATANABE, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | 2D Object recognition system tends to fail when detected features are vastly lost caused by overlapping of multiple objects, occlusion and decline in contrast. This paper proposes a method to manage the incomplete feature detection problem by applying a "relation model" in addition to the conventional object model. First, detection of edge segments, raster-to-vector transformation and supplementation of edge segments are executed and closed region-based descriptions are created. Next, candidates of recognized objects are extracted by calculating the scores of each closed region based on matching with the object model, and unified scores are calculated by unifying the closed region scores in the neighborhood. Then, the consistency among recognized object candidates are examined by applying the relation model, and combinations of closed region are selected as recognized object regions. Finally recognition of unknown regions is executed by applying the relations among object positions and the relations between objects and human motion. Experimental results in indoor scenes by the implemented system are shown. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Object Recognition / Object Model / Relation Model / Model-based Vision |
Paper # | PRMU2006-90 |
Date of Issue |
Conference Information | |
Committee | PRMU |
---|---|
Conference Date | 2006/9/2(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) | Object Recognition using Relation Models |
Sub Title (in English) | |
Keyword(1) | Object Recognition |
Keyword(2) | Object Model |
Keyword(3) | Relation Model |
Keyword(4) | Model-based Vision |
1st Author's Name | Daisuke KATSUMA |
1st Author's Affiliation | Kagoshima University() |
2nd Author's Name | Daisuke SHIMIZU |
2nd Author's Affiliation | Kagoshima University |
3rd Author's Name | Mutsumi WATANABE |
3rd Author's Affiliation | Kagoshima University |
Date | 2006-09-09 |
Paper # | PRMU2006-90 |
Volume (vol) | vol.106 |
Number (no) | 230 |
Page | pp.pp.- |
#Pages | 8 |
Date of Issue |