Abstract
Because of the large and expanding domain of digital imagery, automated segmentation continues to be a challenging task. As a result, recent approaches exploit user input to guide the segmentation process interactively. In this talk I will discuss and contrast three interactive segmentation techniques that make use of live-wire, live-region and live-surface feedback. The live wire metaphor was developed for Intelligent Scissors introduced at SIGGRAPH'95. Intelligent Scissors formulates boundary detection as a user-guided optimal graph search using Dijkstra's algorithm. When the gestured mouse position comes in proximity to an object edge, a live-wire boundary "snaps" to, and wraps around the object of interest. Live region applies a painterly metaphor using min graph cut, implemented on a tobogganed watershed hierarchy, to perform object segmentation and selection. Live surface extends this to three dimensions using a cascaded graph cut to maintain interactivity even when segmenting objects from image volumes.

BIO
Bill Barrett is a Professor of Computer Science at Brigham Young University where he has served as Department Chair and Associate Chair. Barrett received his Bachelor's degree in mathematics and his PhD in Medical Biophysics and Computing, both at the University of Utah. Prior to coming to BYU, he was a research fellow at NIH and worked in private industry for several years, pioneering work in Digital Angiography and 3D Medical Imaging. Current research areas include interactive segmentation techniques, document image analysis and image editing (barrett.cs.byu.edu). Barrett founded the Family History Technology Workshop, now in its 11th year, and also co-founded and co-chaired the First International Workshop on Historical Document Imaging and Processing (HIP'11) held in conjunction with ICDAR2011. Dr. Barrett is best known for his research with Intelligent Scissors, adopted into Adobe Photoshop as Magnetic Lasso.