Data-driven Processing-microstructure relationship identification
While the predictive models from microstructure to material property are developed , the complex mechanism in the material processing procedure hinders the development of our understandings about how the processing conditions will make impacts on the microstructural dispersions in polymer nanocomposites. In this project, starting with the experimental samples, we established the energetics mediated processing-structure relations using statistical dimension reduction algorithms and data mining techniques.
Pixel-wise image processing
Before the microstructure characterization process, grayscale Transmission Electron Microscopic (TEM) and Scanning Electron Microscopic (SEM) images are binarized to two-phase (filler-matrix) images. For polymer nanocomposites with low volume fraction of inclusions, the typically used global thresholding methods will misclassify the image pixels due to even a small portion of shades or uneven background brightness. To address this problem, pixel-wise image processing algorithms are developed to enhave the accuracy of the binarization process.
Active Researcher on the project:
Relevant Publications:
Xiaolin Li, Irene Hassinger, Yichi Zhang, He Zhao, Hongyi Xu,Linda Schadler, L. Catherine Brinson, Wei Chen, A Hybrid Modeling Approach for Establishing Energy-Mediated Processing-Structure Relationship of Polymer Nanocomposites, MRS 2014 Fall Meeting and Exhibit, Boston, USA