Paper
13 August 2010 Exploiting spectral and polarimetric data fusion to enhance target detection performance
Author Affiliations +
Abstract
A multimodal sensor data fusion experiment was performed by exploiting DIRSIG's synthetic data generation capabilities in an urban target detection scenario. Each material in the synthetic scene was attributed with realistic spectral and polarimetric properties, enabling a radiometrically correct calculation of the sensor-reaching-radiance for a notional hyperspectral, multispectral, or polarimetric sensor. The hyperspectral and multispectral data were separately fused with polarimetric data at both the decision and pixel levels, and the impact was assessed by comparing the area under the ROC curves generated by the fused data to the area under the ROC curve generated by the spectral data alone. The impact of additional polarimetric data was shown to be highly dependent on the sensor's viewing geometry, reiterating the complexities involved in polarimetric imaging applications. Also, the impact of additional polarimetric data was demonstrated to depend on the quality of the spatial-spectral information, illustrating the potential to trade spectral resolution for spatial resolution. Further, the decision level fusion algorithm was shown to outperform the pixel level fusion algorithm for the viewing geometries considered, a difference partially explained by the extremely decorrelated nature of the score metrics used as inputs to the decision fusion algorithm.
© (2010) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Brian M. Flusche, Michael G. Gartley, and John R. Schott "Exploiting spectral and polarimetric data fusion to enhance target detection performance", Proc. SPIE 7812, Imaging Spectrometry XV, 78120C (13 August 2010); https://doi.org/10.1117/12.860640
Lens.org Logo
CITATIONS
Cited by 3 scholarly publications.
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Polarimetry

Sensors

Detection and tracking algorithms

Target detection

Image sensors

Data fusion

Image analysis

RELATED CONTENT


Back to Top