ML20005A865
| ML20005A865 | |
| Person / Time | |
|---|---|
| Site: | University of Michigan |
| Issue date: | 06/17/1981 |
| From: | Ganapathy S MICHIGAN, UNIV. OF, ANN ARBOR, MI |
| To: | |
| Shared Package | |
| ML20005A863 | List: |
| References | |
| NUDOCS 8107010400 | |
| Download: ML20005A865 (2) | |
Text
. _ _ _ - _ _ _ _ _ _ _
~
n k-)
ULTRASONIC IMAGING LABORATORY Monthly Report for May 1981 Report Submitted to Dr. Joseph Muscara Nuclear Regulatory Commission by Prof. S.
Gr.napathy REAL TIME PROCESSING Several discussions were held concerning the development of a real time SAFT system.
We are currently looking at both processor and memory speed per poi
-equirements, based on such parameters as frequency, included angle, depth in material, etc.
These figures should give an upper bound on memory access time, and dictate the type of memory structure coquired for different choices of parameters.
The idea of "Sof tware Machining" to adjust for irregular surfaces was also investigated, showing that the error involved is small.
" Software Machining" involves discarding data at the top of A-scans to bring the first points of all A-scans to the same level.
i GRAPHICAL DISPLAY OF FLAWS The Perspective Plotting software has been installed on the VAX and has been thoroughly tested.
An analysis of the computational complexity of triangulation algorithns was performed, and it appears I
that the Keppel technique is superior to the Christianson heuristic in terms of the number of computations needed to make a connectivity decision.
8.187o 2 o ydQ egro w %,es.we.g er g mye e,
g> w m
,es.g
~
p page 2 Monthly Report for May 1981 FLAW DETECTION To improve the flaw detector, work was continued'on the charac-
' terization of grain noise.
A simple estimator which finds the autocorrelation function of the grain noise by averaging the local correlations of a set of A-scans seems sufficient; therefore, this type of estimator will be implemented.
It was also decided to perform all filtering and detection in the time domain, to eliminate some of the problems associated with using i.T processing, such as the need for assuming periodicity.
AUTOMATIC BOUNDARY EXTRACTION Several papers about the Hueckel operator 'iere read, and it appears to be more than enough for our ultrasonic images.
Such an operator would be appropriate for noisy mages or ones with textured background.
Several papers comparing various edge detectors were also read, but they do not seem to do a very thorough comparison.
SPOTLIGHT MODL IMAGING A dissertation was read in order to investigate the advantages of spotlight mede imaging over conventional sesnning.
A detailed and comprehensive assessment will be forthcoming.
SYSTEM UPGRADE The computer room renovation seems to be going on schedule and most of the equipment has arrived.
The new data acquisition system is also nearly ready.
6/17/81 msw
_e w.