| Project/Area Number |
15K06055
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Communication/Network engineering
|
|---|
| Research Institution | Tokyo Institute of Technology |
|---|
Principal Investigator |
Ito Izumi 東京工業大学, 工学院, 助教 (90583435)
|
|---|
| Research Collaborator |
Egiazarian Karen
|
|---|
| Project Period (FY) |
2015-04-01 – 2019-03-31
|
| Project Status |
Completed (Fiscal Year 2018)
|
| Budget Amount *help |
¥4,290,000 (Direct Cost: ¥3,300,000、Indirect Cost: ¥990,000)
Fiscal Year 2017: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
Fiscal Year 2016: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
Fiscal Year 2015: ¥1,950,000 (Direct Cost: ¥1,500,000、Indirect Cost: ¥450,000)
|
|---|
| Keywords | 画像マッチング / 離散コサイン変換 / ハール変換 / 大型変換 / 大型画像変換 / 整数変換 / 直交変換 |
|---|
| Outline of Final Research Achievements |
Discrete cosine transform approximation with large size and Haar transform with arbitrary size were designed for image matching and their accuracy in image matching were evaluated. In the discrete cosine transform approximation based on matrix factorization algorithms, the computational complexity is reduced was due to the architecture simplified by substituting floating-point multipliers with integer multipliers and shift operators, and the accuracy of image matching is almost the same as the conventional one. In the Haar transform with arbitrary size, the mathematical expression for basis images was derived designing for the efficient use of an integral image. Keeping the accuracy of the full search method, the transform reduces the computational load drastically.
|
| Academic Significance and Societal Importance of the Research Achievements |
これまで，ハードウェア構成を考慮した離散コサイン変換は，主に8x8のサイズに限定され，ハール変換は適用サイズが2の冪乗に限られていた．画像の解像度が飛躍的に増大する今日では，より大きなサイズに適応した，計算コストおよびメモリ使用量を抑制した変換が必要となる．今回開発した近似コサイン変換，および任意サイズハール変換は，高い画像マッチング機能を有しながらも，計算量を大幅に削減可能であり，安価なデバイスにおいても使用可能である．
|
