16 September 1987 Synthetic Discriminant Functions For Target Correlation
Author Affiliations +
Abstract
Synthetic discriminant functions (SDFs) are used to construct digital filters for identifying and registering tank targets on 8-bit digitized forward, looking infrared (FLIR) imagery through correlation. The synthesis and testing of SDFs which control the shape of the correlation surface about the peak are discussed. This is effected by shifting the reference images in the + x and ± y directions and encoding these shifted images onto the filter. The advantages of shifting the reference images include sharper target correlation peaks, suppression of sidelobes on the correlation surface, correlation peaks that occur closer to the registration point of the target, and better overall target discrimination. The reference and test imagery consist of aspect-varying tank images, with a high degree of clutter resident on the test set imagery. Successive filter reference images are chosen on the basis of being quantitatively most dissimilar, in a correlation sense, than other images already resident on the filter. Filter construction techniques are discussed, and train and test set classification results are presented.
© (1987) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Edward L. Jenkins, Edward L. Jenkins, John W. Morris, John W. Morris, S. Richard, S. Richard, F. Sims, F. Sims, } "Synthetic Discriminant Functions For Target Correlation", Proc. SPIE 0781, Infrared Image Processing and Enhancement, (16 September 1987); doi: 10.1117/12.940545; https://doi.org/10.1117/12.940545
PROCEEDINGS
8 PAGES


SHARE
Back to Top