September 21, 2011
Speaker: Dr. Michael Greenspan, Dept. of Electrical and Computer Engineering, Queen's University.
Title: Object Recognition in Range Data
Abstract: In this talk, I will review the progress that my students and I have made since 2001 in the field of range image vision. The motivation of our research has been the automatic robotic grasping of free-flying satellites. This application is particularly challenging due to the sparseness of the data that is acquired by space-qualified range vision sensors, which may comprise only thousands or hundreds of points per frame. I will focus on two model-based algorithms for pose determination and object recognition in range data. One is a local method, called Variable-Dimensional Local Shape Descriptors, that automatically selects the best feature properties, using a training phase. The second is a global method, called Potential Well Space Embedding, that exploits the existence of local minima of the Iterative Closest Point potential well space. I will describe and compare these two algorithms, and describe future research directions, as well as a number of interesting applications.
Biography: Michael Greenspan is a Professor within and Head of the Department of Electrical and Computer Engineering, with a cross-appointment to the School of Computing at Queen's University, Kingston, Canada. His research interests include pose determination, object recognition, and tracking (especially using range data), and robotic gaming systems. Professor Greenspan was the recipient of the Canadian Image Processing and Pattern Recognition 2003 Young Investigators Award, and the Premier's Research Excellence Award.
Title: Object Recognition in Range Data
Abstract: In this talk, I will review the progress that my students and I have made since 2001 in the field of range image vision. The motivation of our research has been the automatic robotic grasping of free-flying satellites. This application is particularly challenging due to the sparseness of the data that is acquired by space-qualified range vision sensors, which may comprise only thousands or hundreds of points per frame. I will focus on two model-based algorithms for pose determination and object recognition in range data. One is a local method, called Variable-Dimensional Local Shape Descriptors, that automatically selects the best feature properties, using a training phase. The second is a global method, called Potential Well Space Embedding, that exploits the existence of local minima of the Iterative Closest Point potential well space. I will describe and compare these two algorithms, and describe future research directions, as well as a number of interesting applications.
Biography: Michael Greenspan is a Professor within and Head of the Department of Electrical and Computer Engineering, with a cross-appointment to the School of Computing at Queen's University, Kingston, Canada. His research interests include pose determination, object recognition, and tracking (especially using range data), and robotic gaming systems. Professor Greenspan was the recipient of the Canadian Image Processing and Pattern Recognition 2003 Young Investigators Award, and the Premier's Research Excellence Award.