| Presentation | 2005-06-16 Fast image retrieval based on saliency-weighted interest point matching Akisato KIMURA, Takahito KAWANISHI, Kazuhiro OHTSUKA, Kunio KASHINO, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | We present a new method for content-based image retrieval (CBIR). Region-based approach would be effective to tackle with various kinds of fluctuation such as scale, viewpoint, and foreground-background changes. Many algorithms to extract interest points or regions have been proposed for removing such fluctuations. In this report, we propose a new concept of CBIR. The proposed method utilizes saliency for weighting, not for extracting interest regions. That is, interest points are weighted by neighboring saliency. Saliency-weighted sililarity measure is used for feature matching. One of the problems in the proposed framework is that applying weighted similarity to existing indexing methods is difficult. We also propose a new indexing method called two-stage index search. The first stage is based on saliency and the second one is based on (non-weighted) similarity. Two-stage index search enables us to quickly detect all point features such that saliency-weighted similariry value exceeds a predefined threshold. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | content-based image retrieval / interest points / saliency / weighted silimarity / two-stage index search |
| Paper # | DE2005-2,PRMU2005-23 |
| Date of Issue |
| Conference Information | |
| Committee | PRMU |
|---|---|
| Conference Date | 2005/6/9(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 | Pattern Recognition and Media Understanding (PRMU) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Fast image retrieval based on saliency-weighted interest point matching |
| Sub Title (in English) | |
| Keyword(1) | content-based image retrieval |
| Keyword(2) | interest points |
| Keyword(3) | saliency |
| Keyword(4) | weighted silimarity |
| Keyword(5) | two-stage index search |
| 1st Author's Name | Akisato KIMURA |
| 1st Author's Affiliation | NTT Communication Science Laboratories, NTT Corporation() |
| 2nd Author's Name | Takahito KAWANISHI |
| 2nd Author's Affiliation | Plala Networks Inc. |
| 3rd Author's Name | Kazuhiro OHTSUKA |
| 3rd Author's Affiliation | NTT Communication Science Laboratories, NTT Corporation |
| 4th Author's Name | Kunio KASHINO |
| 4th Author's Affiliation | NTT Communication Science Laboratories, NTT Corporation |
| Date | 2005-06-16 |
| Paper # | DE2005-2,PRMU2005-23 |
| Volume (vol) | vol.105 |
| Number (no) | 118 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |