Detecting a Target of Unknown Brightness in Clutter.
Abstract
Standard infrared target detection algorithms based on the Bayes decision rule or Neyman-Pearson test are optimal only when testing for targets of known strength (brightness). The discriminant functions used in these tests depend, in functional form, on the assumed brightness of the targets being sought. That is to say, they are not uniform with respect to target brightness. Linear uniform tests such as the matched filter are not near optimal for multidimensional cases. The case of interest here is a multidimensional one-the detection of moving targets in differenced mosaic images. The uniform tests that we consider is the generalized maximum likelihood (GML) tests. Three implementations are discussed. Results are presented that indicate that the uniform GML test compares favorably with the optimal Bayes decision rule for detection of moving targets in mosaic imagery.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 27, 1987
- Accession Number
- ADA180925
Entities
People
- Charles F. Osgood
- Richard G. Priest
Organizations
- United States Naval Research Laboratory