ECE Seminar: Analysis of the Cosparse Model for Signals and Images

May 17

Friday, May 17, 2013

11:45 am - 1:00 pm
Fitzpatrick Center Schiciano Auditorium Side B

Presenter

Raja Giryes, Department of Computer Science, Technion

The cosparse analysis model has been introduced recently as an interesting alternative to the standard sparse synthesis approach. In this talk I will point to the differences between the two models and the advantages and disadvantages that the analysis framework introduce for signal and image processing. A general recipe for generating analysis algorithms from existing synthesis ones will be presented, together with theoretical guarantees for several such "converted techniques". The impact of these results on the sparse synthesis framework will be discussed as well, highlighting a new view to what seems to be already classical results in sparse approximation theory. I will conclude with some open problems still unanswered. BIO: Raja Giryes is studying at the Technion, Israel Institute of Technology, for his PhD in Computer Science. He is researching stability and other properties of sparse representations in the synthesis and analysis framework. His main focus is on the theoretical and practical aspects of 'greedy-like' type of algorithms. Raja's research has many applications in the field of image processing and computer vision. He hopes to provide additional insights into existing analytical methods, and to find a solution to the problem of instability, that could lead to higher quality and more effective computer imaging. His faculty advisor is Prof. Michael Elad. Host: Prof. Guillermo Sapiro

Contact

McLain, Paul
660-5254
paul.mclain@duke.edu