Unveiling the Interior of 3D Micro-CT from Iron-Sand Composite
Among the options for reducing the amount of carbon dioxide in our atmosphere is "carbon sequestration," in which greenhouse gases would be confined underground. But before such a plan is implemented, researchers need to understand the properties of the materials surrounding the gases and how those surfaces and gases would interact. This project is aimed at designing nondestructive techniques for quantifying the features of these porous materials and addressing their 3D geometries for carbon sequestration. The team's strategy, illustrated in this movie, is to use computer vision techniques to remove image "artifacts" inherent in using computer tomography to gather the data. Bilateral filters are used to minimize the spurious heterogeneity of the image. Then the data points are classified using a method known as statistical region merging. A statistical test is used to determine the pertinence of a certain pixel/region to another region based on analysis of mean intensities, similarities and the deviation of observed differences between separate regions of the image.
Scientific research by Peter Nico of Berkeley Lab's Earth Sciences Division. Image processing by Dani Ushizima of Berkeley Lab's Computational Research Division. Visualizations were computed on NERSC's CrayXT4 "Franklin" system.