| Project/Area Number |
18K11356
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 61010:Perceptual information processing-related
|
|---|
| Research Institution | Japan Women's University (2020-2023)
Okayama University (2018-2019) |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
右田 剛史 岡山大学, 自然科学研究科, 助教 (90362954)
|
|---|
| Project Period (FY) |
2018-04-01 – 2024-03-31
|
| Project Status |
Completed (Fiscal Year 2023)
|
| Budget Amount *help |
¥4,160,000 (Direct Cost: ¥3,200,000、Indirect Cost: ¥960,000)
Fiscal Year 2020: ¥520,000 (Direct Cost: ¥400,000、Indirect Cost: ¥120,000)
Fiscal Year 2019: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Fiscal Year 2018: ¥2,340,000 (Direct Cost: ¥1,800,000、Indirect Cost: ¥540,000)
|
|---|
| Keywords | 共分散行列 / 加重和制約付き加重方程式 / 白色化 / 射影係数 / 外れ値処理 / 照明適応 / ３次元モデル生成 / 姿勢・形状同時推定 / 標準射影係数 / パターン認識 / 理論化 / 顔追跡・認識融合系 |
|---|
| Outline of Final Research Achievements |
(1) Framework of general-purpose pattern recognition is investigated by analyzing a constrained optimization problem of weight equations. In underdetermined cases, optimal solution is reduced to inner product of whitened vectors, and adaptive dimensional selection and thresholding after whitening are effective for improving the recognition rate. In overdetermined cases, optimal solution is also reduced to whitening in another formulation and thresholding after whitening is still effective for improving the recognition rate.
(2) For the face tracking and recognition, 3d modeling of face is important for effective tracking and recognition. A 3d modeling method is developed for single-view face image sequence, and effective modeling is performed when face poses are not biased in the face image sequence. Generalization of this method is desired in future work.
|
| Academic Significance and Societal Importance of the Research Achievements |
加重和制約付き加重方程式は，登録パターンの加重和で未知パターンを表すための線形連立方程式であり，特徴空間の次元数と登録パターン数の大小関係で不足決定系にも過剰決定系にもなる。不足決定系においては，この方程式の（数学的な）最適解が白色化後の内積に帰着できる。一方，白色化後に外れ値処理を導入することにより最適性はなくなるが，有効な認識系を構成できる。特徴空間を小さな次元数に分解することによっても，この特性を利用して認識系を構成できることから，逆伝播学習に代わる認識系の構成論になる可能性があると考えられる。また，白色化後の外れ値処理は，様々なデータ処理に適用可能と考えられる。
|
