From Pictures to Insights

March 14, 2007
2:50 pm - 4:00 pm
Halligan 111
Speaker: Hanspeter Pfister, MERL
Host: Sarah Frisken

Abstract

As the old saying goes, a picture is worth a thousand words. But finding the right words to describe a thousand pictures is very difficult. Today, our ability to acquire image and video data far outstrips our ability to make sense of that data. This is particularly true in the sciences, where huge amounts of image data are acquired from scanners, microscopes, telescopes, and various other instruments. For the past 20 years, visualization has focused on generating images from data. What is now needed are better methods to generate abstractions that help us to gain insights into large collections of image data.

In this talk I will make the argument that modern visualization approaches require hierarchical data analysis to transform the data into meaningful, perceptually intuitive representations. I will discuss a number of data-driven hierarchical representations that tame the complexity of high-dimensional visual data. First, I will address the representation of spatially-varying appearance using a tree-structured factorization method and a new matrix decomposition algorithm. Then I will show how to generalize these ideas to decompose time-lapse video into simple and intuitive components that can be edited. Next, I will discuss the MERL face-scanning project, where we collected a database of over 400 subjects with thousands of images each in order to build high-quality statistical models of human faces. Finally, I will briefly describe the Connectome, an ongoing project with the Harvard Center for Brain Science, which aims to determine the detailed neural circuitry of the brain from nanoscale-resolution images.

Biography

Hanspeter Pfister a senior research scientist at Mitsubishi Electric Research Laboratories and a visiting scholar at Harvard University. His research is at the intersection of visualization, computer graphics, and computer vision. His work spans a range of topics, including point-based graphics, appearance modeling, face recognition, and computational photography. He is the chief architect of VolumePro, Mitsubishi Electric's real-time volume rendering hardware for PCs. Dr. Pfister received his PhD in Computer Science in 1996 from the State University of New York at Stony Brook and his MS in Electrical Engineering from ETH Zurich, Switzerland, in 1991. He is an instructor at the Harvard Extension school and has taught courses at major graphics conferences including ACM SIGGRAPH, IEEE Visualization, and Eurographics. Dr. Pfister is chair of the IEEE Visualization and Graphics Technical Committee (VGTC) and co- editor of the 2006 NIH/NSF Visualization Research Challenges report. He is a senior member of the IEEE Computer Society and member of ACM, ACM SIGGRAPH, and the Eurographics Association. You can contact him at pfister@merl.com.