2004 Fiscal Year Final Research Report Summary
High-Accuracy Image Interpolation Using Multi-dimensional Variable Fractional-Delay Digital Filters
| Project/Area Number |
14550379
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
情報通信工学
|
|---|
| Research Institution | Toho University |
|---|
Principal Investigator |
ITO Noboru Toho University, Faculty of Science, Associate Professor, 理学部, 助教授 (00237041)
|
|---|
| Project Period (FY) |
2002 – 2004
|
| Keywords | digital filter / Variable digital filter / fractional-delay filter / image interpolation / resolution conversion / Lagrange interpolation / one-dimensional VFD filter / two-dimensional VFD filter |
|---|
| Research Abstract |
During this 3-year research, we focused on developing optimal one-dimensional (1-D) and two-dimensional (2-D) variable fractional-delay(VFD) digital filters for high-resolution image interpolation. Since 2-D VFD filters can be constructed by cascading a pair of 1-D VFD filters with different delay elements, so developing new algorithms for designing and implementing 1-D VFD filters is particularly important, which is the basic step for obtaining good 2-D VFD filters. Therefore, we first developed several new methods for designing and implementing 1-D VFD filters, then generalized the 1-D VFD filters to the 2-D case for high-accuracy image interpolation. The new techniques developed during this research are as follows :
(1)weighted-least-squares method for designing VFD filters with also tunable magnitude ;
(2)SVD-based design and new structures for VFD filters ;
(3)VFD filter design using vector-array decomposition ;
(4)VFD filter design using coefficient-symmetry ;
(5)theoretical proof of Lagrange-type VFD filters ;
(6)2-D Lagrange-type VFD filters for high-resolution image interpolation.
We have used many image examples to show that the image interpolation using 2-D VFD filter is superior to other existing image interpolation techniques in terms of high-accuracy and reduced computational complexity.
|
