1993 Fiscal Year Final Research Report Summary
Optimization of Sparce-Data Computed Tomography
| Project/Area Number |
04650384
|
|---|
| Research Category |
Grant-in-Aid for General Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
計測・制御工学
|
|---|
| Research Institution | Toyama Prefectural University |
|---|
Principal Investigator |
IWAMA Naofumi Toyama Kenritsu University, Department of Electronics and informatics, Professor, 工学部, 教授 (30023253)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
HATTORI Katsumi Toyama Kenritsu University, Department of Electronics and informatics, Research, 工学部, 助手 (60244513)
TERANISHI Masaru Toyama Kenritsu University, Department of Electronics and informatics, Research, 工学部, 助手 (50237004)
|
|---|
| Project Period (FY) |
1992 – 1993
|
| Keywords | Computed tomography / Sparce data / AIC / GCV / Series expansion / Linear and nonlinear regularizations / Plasma and ion beam imagings / Velocity distribution recovery |
|---|
| Research Abstract |
The computed tomography with sparce data has been studied from a standpoint of optimizing the image reconstruction in application to the followings :
(1) Electron velocity distribution recovery of low temperature plasma with the Langmuir probe ;
(2) Tomography of the density distribution of heavy ion beam with helical metal wires ;
(3) Visible light emission tomography of plasma in the poloidal plane of a small tokamak ;
(4) Soft X ray emission tomography of plasma in the superconducting tokamak Tore Supra.
Conclusions :
1.The ART is weak in regularization but good enough for the velocity distribution when the projection sampling is fine. The resolution is high enough for the point sources of heavy ion if 3 parallel beam projections are available.
2.The Phillips-Tikhonov(PT)method is linear with a strong regularization, which can be optimized with the neg-entropy GCV.The behavior of GCV are stable even for the data missing in toroidal plasma imaging.
3.The SVD method is also linear and assisted with the GCV, but the decomposition is not so reasonable as that pf Foureir and is too weak in regularization for toroidal plasma imaging.
4.The maximum entropy method is nonlinear, good is resolution and weak in regularization compared with the PT method. Computing time reductoion and criterion for optimization are wanted.
5.The series expansion method is linear with the information compression, which can be optimized with the neg-entropy AIC.On the MHD oscillations of the Tore Supra plasma, the empirically best term number has been supported with the minimum of AIC.
|
