Image analysis – Pattern recognition – Feature extraction
Reexamination Certificate
2001-08-14
2004-11-02
Dastouri, Mehrdad (Department: 2623)
Image analysis
Pattern recognition
Feature extraction
C382S203000, C382S218000, C382S170000
Reexamination Certificate
active
06813380
ABSTRACT:
BACKGROUND OF THE INVENTION
A known procedure for target detection involves hyperspectral line pairs wherein two images of the same target area are obtained using two different filters of different wavelengths. With the correct selection of filters, the images may be subtracted, resulting in an enhanced target signature.
By predetermining hyperspectral lines, or wavelengths, which will be interrogated for a target of interest under known battlefield conditions, it would be possible to decrease the computational burden associated with data acquisition utilizing a hyperspectral sensor imager. The present invention provides a method for meeting this objective.
SUMMARY OF THE INVENTION
The method of the present invention determines hyperspectral line pairs which will indicate a target of interest in a target area. Images are taken of a known target in a target area through two different wavelength filters, constituting a hyperspectral line pair, and the obtained images are subtracted, yielding a resultant image. Values are obtained of pixels of the resultant image along vertical and horizontal lines which pass through the target to obtain Y and X profiles of pixel values.
A first of the profiles is analyzed to extract peak signal, peak noise and mean noise values within the profile and these are examined to see if a predetermined relationship exists between these values. If not, a new hyperspectral line pair is selected to obtain a new resultant image for analysis. If the relationship does exist, first and second user defined metrics are created, relating to the difference in predetermined ones of the peak signal, peak noise and mean noise values.
The first and second user defined metrics are tested to see if both exceed predetermined respective first and second thresholds, and if not, a new hyperspectral line pair is selected to obtain a new resultant image for analysis. If both the thresholds are exceeded, a third metric is created which is a ratio of the first and second metrics.
This third metric is tested to see if it exceeds a predetermined third threshold, and if not, a new hyperspectral line pair is selected to obtain a new resultant image for analysis, and if the third threshold is exceeded, the above steps are performed on the second of the profiles. Hyperspectral line pairs which result in the third threshold being exceeded for both the Y and X profile analysis are selected as candidates for positive target detection, with the potential for target identification.
REFERENCES:
patent: 6282301 (2001-08-01), Haskett
patent: 6490075 (2002-12-01), Scheps et al.
patent: 6529614 (2003-03-01), Chao et al.
patent: 6751359 (2004-06-01), Handley
patent: 2002/0015151 (2002-02-01), Gorin
1998 Thesis Abstract; United States Army, Navy, Air Force; pp. 35-53.
Kovel Steven
Sola Marcos C.
Adams William V.
Dastouri Mehrdad
The United States of America as represented by the Secretary of
LandOfFree
Method of determining hyperspectral line pairs for target... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of determining hyperspectral line pairs for target..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of determining hyperspectral line pairs for target... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3310429