Performance-driven Multimodality Sensor Fusion
Abstract
The broad objective of this grant was to develop a generally applicable theory of performance of information-level fusion that provides accurate prediction of post-fusion algorithm accuracy in uncertain environments. determines factors affecting fundamental performance tradeoffs, e.g., sample size, resolution, specificity, and sensitivity of sensors. specifies performance benchmarks allowing quantitative comparison of different fusion algorithms. provides guidelines for algorithm design and optimization. The effort focused on information theoretic fusion methods and our analysis was based on geometric properties of information. Our research has impacted application domains where information theoretic fusion is applied. These included georegistration, remote sensing, multimodality anomaly detection, visualization, and dimensionality reduction.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 23, 2012
- Accession Number
- ADA565491
Entities
People
- Alfred O. Hero III
- Raviv Raich
Organizations
- University of Michigan