Due: Saturday, April 28, 2018 before 12:00 pm
While the objectives of the final projects are specific to each selected topic, their overarching goal is to tackle a non-trivial visualization problem and thereby devise, implement, and demonstrate an effective visualization strategy on an interesting dataset. In particular, deciding what particular visualization algorithm(s) to use and combine is part of each project's tasks.
Your submission for the final project must include:
Below is a partial list of themes and datasets available for the completion of the final projects. As indicated previously, you are more than welcome to use other datasets (e.g., from your own research), provided this has been discussed and agreed upon with the instructor.
For those of you interested in visualizing the human anatomy, the Visible Human Project (National Library of Medicine, NIH) offers an invaluable resource. Yet, accessing the data and formating it to your needs can be cumbersome and time consuming. To facilitate your task, I have prepared some datasets that can be readily visualized with VTK.
CT scans for male and female patients can be downloaded from this URL. As mentioned in class, the datasets are provided in DICOM format, which can be imported with various tools. Paraview in particular allows you to import an entire DICOM directory and to export it to VTK format.
MRI datasets corresponding to T1-weighted, T2-weighted, and proton density images (see this page for explanations) of the male patient's full body can be downloaded on the class web page. Following files are available.
Note that the individual portions listed above of the visible male MRI scan overlap and together form a full body scan. However, you will need to translate and align them properly to obtain the correct result.
We saw in class that diffusion tensor MRI (or DTI) constitutes a typical source of tensor field visualization data with important applications in the biomedical study of fibrous tissues such as the brain's white matter or the heart. The datasets below cover these two scenarios. In each case, the diffusion tensor field is accompanied by a scalar mask that indicates which portions of the volume contain meaningful signal values, a scalar field corresponding to the fractional anisotropy (FA), and three additional scalar datasets containing the eigenvalues.
Two human brain DTI scans and a synthetic DTI dataset are available.
The first dataset is a DTI scan that was part of the IEEE Visualization Challenge 2010 and is courtesy of Prof. B. Terwey, Klinikum Mitte, Bremen, Germany.
The second dataset is a human brain scan of a healthy patient acquired at 1.6mm x 1.6mm x 1.5mm resolution. It was first analyzed as part of this paper.
The third dataset is courtesy of the IIT Human Brain Atlas. It corresponds to a representative average DTI dataset against which individual patients' scans can be registered.
The first dataset was produced as part of a PhD thesis by Nicolas Toussaint at INRIA in France. The dataset is high resolution and unusually clean, since it is the result of a image-based modeling procedure.
The second dataset corresponds to a canine heart. This dataset is more clinically relevant but also significantly noisier than the previous one.
Three datasets are available for projects focused on flow visualization for CFD analysis. Two of these datasets correspond to time-independent (so-called Reynold Averaged Navier-Stokes) simulations over a three-dimensional unstructured and adaptively refined mesh. The third dataset in contrast corresponds to a time-dependent simulation over a two dimensional domain defined by a uniform grid.
This dataset consists of the flow past a large sedan car. A quick inspection of the dataset will show you that the body of the vehicle occupies a tiny portion of the overall computational domain. This dataset is courtesy of Dr. Markus Ruetten, Deutsche Luft- und Raumfahrt, Goettingen, Germany.
Dataset available here: 8.6M vertices, 26.9M cells, 670 MB.
The second data set shows the flow around the (half) body of an airliner that contains a single turbine. In this case, the body of the plane lies on one of the boundaries of the domain. This dataset is also courtesy of Dr. Markus Ruetten, DLR Goettingen, Germany.
Dataset available here: 7.46M vertices, 8.58M cells, 654 MB.
The third dataset describes the velocity of the ocean in the Gulf of Mexico over a period of 5 months (April 1 - August 31, 2013). This dataset was produced using the Navy Coastal Oceanic Model (NCOM) by the Naval Research Laboratory, NOAA. The original dataset is huge. The dataset listed below is a spatially and temporally downsampled version of the original.
Dataset available here: 450 x 355 vertices, 12h time resolution, 292 MB (compressed).
Information about the dynamics of the solar system is available from a variety of sources. A convenient web server to access all the necessary data is hosted by the Solar System Dynamics group at NASA/JPL. The web interface can be accessed here. Textures for the realistic rendering of the planets is available at this URL.
Your final project must be submitted before Saturday, April 28, 2018 before 12:00 pm. Refer to the instructions below.
CMakeLists.txt if applicable.#! python2" at the beginning of the file, if you are using Python 2.7, and add "#! python3" if you are using python 3.x. This is to make my grading life easier. (See here for detail on py launcher)
README.txt
file with compilation / execution instructions
(optional).<myLogin>_final, where <myLogin> is your Purdue login.Put all your files in a folder and package your folder to a zip file. Submit your file through Blackboard submit. You can do so by logging into Blackboard, go to Course Content on the left column, and click on the "Assignment: Final Project" link. There will be a section to attach files. Please make sure your file size is less than 100MB. Do not leave your submission to last minute, as often time Blackboard would be overloaded with requests and hang, and you may not get your submission uploaded on-time.