ESM4714
Scientific Visual Data Analysis and Multimedia

Assignment #3: Small Multiples

Due: Two weeks from the day it was assigned

Goals:
Physical Interpretation of multiple CT slices.
Working with 3D data sets and creating different HDF formats and a "pics" and QuickTime animation.

Classroom Handout:
Pick up your handout on PV-Wave procedures and HDF-C/FORTRAN program example on how to make the different types of HDF files.

Part I (80pts):
Create pics, QT and different types of HDF files

Determine the location (in pixels) of three regions where large voids or gradients of density exist in a Computed Tomography (CT) scan of a titanium turbine fan blade. These inhomogeneities are created by a metal casting process. The 3D compressed data file is located on your optical disk in the directory /optical/ESM4714/ examples/kriz/fan.128_ascii.Z . With simple C and FORTRAN programs take all 62 sequential planes ("slices") and create a single pics file and QuickTime Movie, fan_128.pics and fan_128.Moov respectively (store these files on the Mac formatted optical disk located near Mac_Development). Also create a single Raster Image Set (RIS) Hierarchical Data Format (HDF) file, fan_128_ris.hdf for the 30th slice, and a single Scientific Data Set (SDS) HDF file, fan_128_sds.hdf.

Part II (10pts):
Physical limitations of data sets

As CT technology advances we anticipate CT scans that can have many more planes. In this example we will pick an upper limit of 1024 planes where each plane has 1024 by 1024 pixel resolution and each pixel dimension represents a 20 micro meter cube. Material researchers would like to reduce the 3D pixel dimension to a smaller size to resolve smaller cracks.

Such CT scanners exists at several government research facilities. At NASA Langley such scanners are mounted on a mechanical load frame where materials are subject to tension loads and the resulting crack and void coalition can be observed as a "3D distributed damage state" that grows ("evolves") into a critical damage state of fracture. For example with such a system the growth of a damage state can be tracked at twenty different increasing load levels.

Using the pixel resolution state above how much data, in Mbytes, would result from a test with twenty load scans? How much more storage would be required if we doubled the pixel resolution to detect smaller cracks? If the researcher asked for more scans at smaller increments of load, how much would this increase the storage requirements? How would you organize and store these data sets without losing critical physical information for quantitative material characterization (i.e. number of significant figures: integers .vs. floating point)? Are there hardware-software limits for volume visualization of such large data sets? Include these comments in your lastname.txt file.

Part III (10pts):
Post your results on the homework account

Logon to username: homework, password: to be handed out in class. Create a directory with your lastname: (~homework/assign#3/lastname).

Put a copy of any C, FORTRAN, and PV-Wave procedure files you used in this assignment, in your lastname directory. Also put a copy of a text file lastname.txt with your comments on homework assignment#3.

Put a copy of your "pics" and QuickTime animation files on the shared optical disk (Mac formatted) located near Mac_Development. Print a copy of your files and also hand these copies in at class on the due date.

Your grade will be based on your observations and conclusions and on what I see in your image files. Your grade will not be based on your programming skills although brevity, clarity, and meaningful comment statements will be appreciated.

Click image to return to Visualization home page.
Ronald D. Kriz
College of Engineering
Virginia Tech
Revised 01/10/99

http://www.sv.vt.edu/classes/ESM4714/Assign/assign3.html