High Performance and Accurate Change Detection System for HyspIRI Missions

Award Information
Agency:
National Aeronautics and Space Administration
Branch
n/a
Amount:
$124,999.00
Award Year:
2012
Program:
SBIR
Phase:
Phase I
Contract:
NNX12CE41P
Agency Tracking Number:
114239
Solicitation Year:
2011
Solicitation Topic Code:
S6.03
Solicitation Number:
n/a
Small Business Information
Signal Processing, Inc.
MD, Rockville, MD, 20850-3563
Hubzone Owned:
N
Socially and Economically Disadvantaged:
Y
Woman Owned:
Y
Duns:
620282256
Principal Investigator:
Chiman Kwan
Principal Investigator
(240) 505-2641
chiman.kwan@signalpro.net
Business Contact:
Chiman Kwan
Business Official
(240) 505-2641
chiman.kwan@signalpro.net
Research Institution:
Stub




Abstract
We propose novel and high performance change detection algorithms to process HyspIRI data, which have been used for monitoring changes in vegetation, climate, coastal and ocean ecosystems, urban areas, etc. First, we propose a novel hybrid in-scene atmospheric correction (H-ISAC) algorithm, which can compensate for distortion of hyperspectral image characteristics due to atmosphere. Conventional ISAC is applicable only to imagers with wavelengths larger than 1 micro-meter and hence it is not applicable to HyspIRI imager which has a wavelength range of 0.38 to 2.5 micro-meters. Our algorithm is simple to implement and does not require any dark pixels in the images. Second, after images are atmospherically compensated for, we propose a novel change detection algorithm known as MRCD (multiple reference change detection) using multiple images collected in the past. Our algorithm can handle misregistration and parallax issues and hence the change detection results will be more accurate. Third, we propose high performance algorithms to determine the content of the changes. For example, what materials are in the changes and where these materials are distributed. We will also determine if there are any new and unseen materials in the changes. To determine known materials with known signatures in the changes, we propose a fast matched signature identification algorithm called Adaptive Subspace Detector (ASD). We compared ASD with several other tools and found that ASD outperformed other methods. To determine any anomalies, we propose a high performance anomaly detection tool called clustered kernel Reed-Xiaoli (CKRX) algorithm. This tool was recently developed by us, is fast, and can achieve very high anomaly detection rate in hyperspectral images from the Air Force. Fourth, the above tools can be implemented in a parallel processing architecture, in which the computations are distributed to multiple processing cores.

* information listed above is at the time of submission.

Agency Micro-sites

US Flag An Official Website of the United States Government