↑ Return to Past Projects

E.I.M Segmentation

Electronic Imaging for Microscopy – Segmentation Tools Data

Client: U.S. Air Force Research Laboratory - Wright-Patterson AFB


Summary
       The Electronic Imaging for Microscopy project is an initiative designed to modernize data analysis for Materials Science by developing and adapting Signal Processing algorithms towards reducing the human interaction necessary for analysis of digitally acquired image data. Since the development of classical image processing algorithms in the 1970's, many algorithmic approaches have been developed for applications such as medical imaging, homeland security, and computer vision that cannot rely on the intervention of experts to analyze data by hand. Invariably, these algorithms involve (ill-posed) inverse methods that require some prior knowledge of what the structures observed should look like in order to produce reasonable answers. The EIM effort uses Materials Science-specific prior knowledge in order to develop Materials Science-specific algorithms for analysis of microscope-acquired image data to develop realistic models of the structures of the materials examined. These models are useful for subsequent activities such as physics-based forward modeling or developing statistical models of structure. All codes and data are contributed to the greater Materials Science community in order to further the goals of Integrated Computational Materials Engineering as well as reducing the time-to-market for new materials.
       This is a joint effort between the Electronic Imaging and Materials Science communities that is part of the ICMD Tools program at the Materials Directorate of the Air Force Research Laboratory. One of the accomplishments was to push forward the state of the art in expectation-maximization/maximization of the posterior marginals (EM/MPM) segmentation algorithms as it relates to materials systems.

[important]Downloads and Code Availability
All code is available from the BlueQuartz Software Git Repository server.[/important]

The code depends on some small parts of boost and for Parallel versions of the code you will need the latest version of Intel Threading Building Blocks (either commercial or open source). In addition to build the Graphical User Interface version of the code you will need Qt version 4.8.x from the Qt Project. This project uses CMake to configure the sources for building.