2011 Fiscal Year Final Research Report
A study of 3D head geometry estimation based on selective use of 2D image and acoustic features
| Project/Area Number |
21700207
|
|---|
| Research Category |
Grant-in-Aid for Young Scientists (B)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
Perception information processing/Intelligent robotics
|
|---|
| Research Institution | Waseda University |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2009 – 2011
|
| Keywords | 画像特徴量 / 音響特徴量 / 顔特徴データベース / 正準相関分析 / ニューラルネットワーク / 3次元頭部形状推定 |
|---|
| Research Abstract |
We aimed to realize a 3D geometry estimation method based on selective use of 2D image and acoustic features and obtained four results as follows. (1) We constructed a database which contains 109 range scans and 5 Japanese vowel speeches. (2) We developed a method for fast 3D geometry estimation from a face photograph using prior knowledge of 3D face geometries. (3) We developed a framework to estimate 3D head geometry from a vowel speech using a neural network trained using a pair of acoustic features and 3D head geometries. (4) We developed a registration method for non-rigid object to effectively generate a 3D head model from a range scan, and several achievements have published as journals.
|
