Title :
On constrained energy minimization and the partial unmixing of multispectral images
Author_Institution :
Environ. Syst. Sci. Centre, Reading Univ., UK
fDate :
3/1/2002 12:00:00 AM
Abstract :
Constrained energy minimization, a target detection algorithm, sometimes appears to correlate well with fractional abundance. This letter shows that exact correlation with true relative abundance arises when the following conditions hold: underlying mixing is linear, end-members are well separated with respect to random noise, and global abundances are constrained in a particular way
Keywords :
feature extraction; geophysical signal processing; image recognition; minimisation; remote sensing; constrained energy minimization; end-members; fractional abundance; global abundances; hyperspectral data; minimization; multispectral images; partial demixing; random noise; target detection algorithm; underlying mixing; 1f noise; Availability; Hyperspectral imaging; Hyperspectral sensors; Image sensors; Matched filters; Minimization methods; Multispectral imaging; Object detection; Pixel;
Journal_Title :
Geoscience and Remote Sensing, IEEE Transactions on
DOI :
10.1109/TGRS.2002.1000332
