Computer Science
Scientific paper
Jul 1994
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=1994spie.2225..350f&link_type=abstract
Proc. SPIE Vol. 2225, p. 350-359, Infrared Detectors and Focal Plane Arrays III, Eustace L. Dereniak; Robert E. Sampson; Eds.
Computer Science
Scientific paper
As new IR focal plane array (IRFPA) technologies become available, improved methods for coping with array errors must be developed. Traditional methods of nonuniformity correction using simple calibration mode are not adequate to compensate for the inherent nonuniformity and 1/f noise in some arrays. In an effort to compensate for nonuniformity in a HgCdTe IRFPA, and to reduce the effects of 1/f noise over a time interval, a new dynamic neural network (NN) based algorithm was implemented. The algorithm compensates for nonuniformities, and corrects for 1/f noise. A gradient descent algorithm is used with nearest neighbor feedback for training, creating a dynamic model of the IRFPA's gains and offsets, then updating and correcting them continuously. Improvements to the NN include implementation on a IBM 486 computer system, and a close examination of simulated scenes to test the algorithms boundaries. Preliminary designs for a real-time hardware prototype have been developed as well. Simulations were implemented to test the algorithm's ability to correct under a variety of conditions. A wide range of background noise, 1/f noise, object intensities, and background intensities were used. Results indicate that this algorithm can correct efficiently down to the background noise. Our conclusions are that NN based adaptive algorithms will supplement the effectiveness of IRFPA's.
Eden Dayton D.
Fenner Jonathan W.
Simon Solomon H.
No associations
LandOfFree
Design of an IRFPA nonuniformity correction algorithm to be implemented as a real-time hardware prototype does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Design of an IRFPA nonuniformity correction algorithm to be implemented as a real-time hardware prototype, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Design of an IRFPA nonuniformity correction algorithm to be implemented as a real-time hardware prototype will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-825129