Optimal Mass Transport for Statistical Estimation, Image Analysis, Information Geometry, and Control
Abstract
We developed several new directions in the theory and applications of Optimal Mass Transport (OMT). OMT has its origins in civil engineering (Monge 1781) and economics (Kantorovich 1942), but in recent years has increasingly impacted a large number of other fields (probability theory, partial differential equations, physics, meteorology). We have addressed computational aspects of the problem and the need for further expanding the arsenal of computational tools. We considered a wide range of generalizations and insights for the purpose of tackling problems of AFOSR interest. These include matrix-valued statistics and fusion of information, optical flow, controlled active vision, tracking and dynamic textures.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 10, 2017
- Accession Number
- AD1025343
Entities
People
- Allen R. Tannenbaum
- Tryphon T. Georgiou
Organizations
- Regents of the University of Minnesota