NCSI

   

Micro-CT Image Segmentation and Visualizations of Additively Manufactured Polymer Composites


Shodor > NCSI > XSEDE EMPOWER > XSEDE EMPOWER Positions > Micro-CT Image Segmentation and Visualizations of Additively Manufactured Polymer Composites

Status
Completed
Mentor NameJun Li
Mentor's XSEDE AffiliationResearch Allocation
Mentor Has Been in XSEDE Community1-2 years
Project TitleMicro-CT Image Segmentation and Visualizations of Additively Manufactured Polymer Composites
SummaryThis project will investigate high-resolution Micro-CT image segmentation and visualizations of additively manufactured (3D printed) polymer fiber composites. A key challenge in using the X-ray CT imaging technique is to accurately segment material phases in contact so as to faithfully visualize the microstructure morphology, which provides the basis for image-based finite element modeling and analysis.
Job DescriptionThe student will learn and apply open-source ImageJ or TomViz software packages to conduct Micro-CT image segmentation and visualizations.
In particular, the Trainable Weka Segmentation (TWS) machine learning technique will be investigated to enhance image segmentation of material phases over the conventional watershed-based method. X-ray CT images of 3D printed polymer fiber composites will be studied in this project.
Computational ResourcesWe will start this project on a local HPC cluster with a reduced size model to practice image software packages and parallel methods. Then we plan to move to cloud-based XSEDE resources for a large size model. Since the image visualizations and analysis require interactive computing with large I/O and datasets, we will request Stampede 2 for this project. When time allows, we will further conduct performance tests of large model visualizations on Stampede 2.
Contribution to Community
Position TypeIntern
Training PlanTraining will be provided for the student to understand image segmentation/visualization techniques as well as use image process software such as open-source ImageJ or TomViz packages. In particular, both conventional watershed-based method and the TWS machine learning technique for image segmentation will be trained.
The student will also be trained to be familiar with Linux operating system and High-performance computing environment to perform large model parallel visualizations.
Student Prerequisites/Conditions/QualificationsMust have programming skills such as in Matlab or Python to develop user scripts. CAD modeling and FEA skills are desired to transform the images into finite element models.
DurationSemester
Start Date09/01/2020
End Date11/30/2020

Not Logged In. Login
 

( ! ) Notice: Unknown: Server 127.0.0.1 (tcp 21201) failed with: Failed reading line from stream (0) in Unknown on line 0