Detection of Unresolved and Subpixel-Scale Anomalous Objects in Hyperspectral Imagery
Abstract
Major Goals: The basic research objectives of this three-year project include the following: 1. to achieve an understanding of the relationship of the complexity of the hyperspectral image data cube and the degree of object anomaly from its immediate neighborhood that is detectable; 2. to investigate alternative algorithms to autonomously detect and locate subpixel anomalies in hyperspectral imagery; 3. to investigate the utility of applying means to optimize the selection of the spectral bands comprising the image data cubes to improve performance and reduction of computational resources; 4. to develop algorithms that detect and locate a subpixel-scale anomalous object when portions of the object are located in adjacent pixels; 5. to investigate algorithms that detect and locate an anomalous object when portions of the object are located in adjacent pixels and fill fully one or several pixels and/or portions thereof; 6. to investigate the impact upon Probability of Detection and False Alarm Rate resulting from the change in the output image from a focal plane array (FPA) as the optical image incident upon the FPA is shifted by subpixel amounts; 7. to investigate the efficacy of improving anomalous object detection performance if the Heidary-Johnson resolution enhancement algorithm is applied to the image data cube to synthetically enhance the data cube as if physical microdithering of the optical image on the FPA had occurred and 8. to experimentally validate anomalous object detection algorithms.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 08, 2019
- Accession Number
- AD1081705
Entities
People
- Barry Johnson
- Chance Glenn
- Kaveh Heidary
- Kenneth Sartor