| Presentation | 2005-10-28 A Study of Face-Detection Methods by Using Self-Adaptive Segmented Template Matching Shingo Kasaki, Akiyoshi WAKATANI, Hiroshi KADOTA, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Template-matching methods to detect the faces within images have been studied. Two-dimensional Harr-like transform is performed to extract the features on the segmented templates and the images under test, and their correlation is computed. Three candidates were proposed for the feature-correlation procedures between the templates and the images by using : 1) the absolute values, 2) the absolute values plus the orientation and 3) the absolute values plus the orientation with a weak freedom of the mutual position of the segments, which we call 'Self-Adaptive Segment'. The relationship between the correlation intensity and the detection rate or the correctness has been estimated for each method against nearly frontal faces in the images. As a result, it is shown that high-detection rates are realized over wider range of facial pose from the pure frontal view by using the procedure of 3). This procedure could be implemented onto the parallel architecture rather smoothly, so the real-time face detection in the motion pictures would be available by a dedicated parallel processor. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Object Recognition / Face Detection / Template-Matching / Harr-like Transform / Self-Adaptive Template |
| Paper # | PRMU2005-98 |
| Date of Issue |
| Conference Information | |
| Committee | PRMU |
|---|---|
| Conference Date | 2005/10/21(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) | A Study of Face-Detection Methods by Using Self-Adaptive Segmented Template Matching |
| Sub Title (in English) | |
| Keyword(1) | Object Recognition |
| Keyword(2) | Face Detection |
| Keyword(3) | Template-Matching |
| Keyword(4) | Harr-like Transform |
| Keyword(5) | Self-Adaptive Template |
| 1st Author's Name | Shingo Kasaki |
| 1st Author's Affiliation | Faculty of Design, Kyushu University() |
| 2nd Author's Name | Akiyoshi WAKATANI |
| 2nd Author's Affiliation | Faculty of Science and Engineering, Konan University |
| 3rd Author's Name | Hiroshi KADOTA |
| 3rd Author's Affiliation | Faculty of Design, Kyushu University |
| Date | 2005-10-28 |
| Paper # | PRMU2005-98 |
| Volume (vol) | vol.105 |
| Number (no) | 375 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |