DSP seminar, Wed Oct. 05 Time and location: 4:00-5:00 PM 2269 Beckman Institute

Title: Surfacelets: Exploring Higher Dimensions with Fewer Tears

Speaker: Yue Lu

Graduate Student, UIUC

Abstract:

High-resolution 3-D and even higher dimensional volumetric data are increasingly 
available in a wide gamut of scientific and technological disciplines ranging from 
biomedical sciences to extragalactic astronomy. 
To efficiently analyze and make sense of such huge amount of data, we need to 
create and employ new tools from various fields of engineering, including signal 
processing. In this talk, I will introduce a new tool for signal representation, 
called surfacelets, that can be used to capture and represent signal singularities 
lying on smooth manifolds of co-dimension 1 (e.g. surfaces in R^3). Such singularities 
are often observed in 3-D medical signals and image sequences (video), where the 
signals are mostly smooth except on some boundary surfaces. The proposed surfacelets 
constitute a tight frame in R^N, and can be efficiently implemented by a tree-structured 
filter bank. In the talk, I will also show a few potential applications of the surfacelets, 
including denoising on 3-D medical data and video sequences. Preliminary results suggest 
that the proposed surfacelets can significantly outperform other more traditional methods.