W&M Homepage

Location

Jones Hall, Room 301
200 Ukrop Way
Williamsburg, VA 23185Map this location

Abstract: Computational topology is a new and vastly growing branch of applied mathematics and computer science. This area, in particular the persistent homology, has been proven to be successful in many applications. Thresholding, one of the most fundamental methods in the image segmentation, can be used to create binary images from a grayscale image. An automated image thresholding method based on the persistent homology is presented. The primary difference among traditional methods is that the resultant binary image respects underlying topological features. Furthermore, in the presence of noise, the method provides more information to obtain a better estimate of the Betti numbers. Finally, we will apply it to firn data, a type of ice, from Climatology.