Advanced Metrology Lab, Department of Industrial & Systems Engineering.
Nanoparticle Detection from Noisy TEM Images: Proposed a robust and accurate algorithm to detect and characterize nanoparticles from noisy transmission electron microscope (TEM) images, by combining the intensity and gradient-based image information. Produced Matlab codes to implement the algorithm efficiently.
Phase Change Detection in Nanocrystal Growth: Proposed an automatic tool to analyze a nanocrystal growth process from in situ TEM videos retrospectively. The proposed method first captures time-varying particle size distributions to depict the dynamic process, then identifies the growth phases driven by different mechanisms, and finally builds a hybrid model for the multi-phase growth.
Online Monitoring via In Situ TEM Videos: Proposed a monitoring method via in situ TEM videos to analyze a nanocrystal growth process prospectively. The proposed method can online characterize and track the current growth status of nanocrystals through state space modeling and modified Kalman filtering, and will trigger an out-of-control alarm when necessary.
Broadband Network & Digital Media Lab, Department of Automation.
Multiview Stereo Reconstruction: Produced Matlab code of majorization-minimization optimization strategy for corrupted matrix recovery from the perspective of compressive sensing. The proposed algorithm can fuse noisy depth maps to construct a 3D model for multiview stereo reconstruction.
Image Understanding and Categorization: Proposed a novel codeword assignment algorithm from the perspective of manifold learning that breaks the restrictions of typical Euclidean distance. The proposed algorithm can improve the accuracy of natural image understanding and synthetic-aperture radar (SAR) image identification.