New independent component analysis method using higher order statistics with application to remote sensing images
Xiaohui Zhang · Optical Engineering · 2002
A new independent component analysis (ICA) method is pro- posed and a new ICA algorithm making use of the higher order statistics is introduced accordingly. We call it the joint cumulant ICA (JC-ICA) al- gorithm. Its application in synthetic aperture radar (SAR) and airborne visible infrared imaging spectrometer (AVIRIS) images are discussed. The results show the potential usage in image processing problems. © 2002 Society of Photo-Optical Instrumentation Engineers. (DOI: 10.1117/1.1482722) Subject terms: higher-order statistics; independent component analysis; penalty function; gradient descent; neural networks; remote sensing.