Performance Metrics for the Evaluation of Hyperspectral Chemical Identification Systems
Abstract
Remote sensing of chemical vapor plumes is a difficult but important task with many military and civilian applications.Hyperspectral sensors operating in the long wave infrared (LWIR) regime have well demonstrated detection capabilities. However, the identification of a plumes chemical constituents, based on a chemical library, is a multiple hypothesis testing problem which standard detection metrics do not fully describe. We propose using an additional performance metric for identification based on the so-called Dice index. Our approach partitions and weights a confusion matrix to develop both the standard detection metrics and identification metric. Using the proposed metrics, we demonstrate that the intuitive system design of a detector bank followed by an identifier is indeed justified when incorporating performance information beyond the standard detection metrics.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 10, 2016
- Accession Number
- AD1034538
Entities
People
- Dimitris G. Manolakis
- Eric Truslow
- Steven E. Golowich
- Vinay K. Ingle
Organizations
- MIT Lincoln Laboratory
- Northeastern University