Location
Salt Lake Community College
Start Date
5-5-2003 9:40 AM
Description
one of the most challenging areas in data fusion is efficient abstraction of relevant information from data sets. In remote sensing, this is particularly important due to the amount of detail that is desired from relatively noisy measurements of sensors with limited capability. The goal is to identify multiple materials that are embedded in a particular pixel and thus, provide for more effective image fusion. However, without prior knowledge of a scene, it is difficult, if not impossible, to identify and classify the number of separate sources in a scene of interest. To address this issue, a method of using independent component analysis (lCA) techniques to solve the unsupervised blind source separation (BSS) problem on hyperspectral data was implemented. This general methodology offers the capacity to determine what endmembers are present in a scene and their levels of existence within individual pixels. The FastICA algorithm was used in conjunction with A VIRIS hyperspectral data to demonstrate this method.
Processing for Image Fusion Using Independent Componen
Salt Lake Community College
one of the most challenging areas in data fusion is efficient abstraction of relevant information from data sets. In remote sensing, this is particularly important due to the amount of detail that is desired from relatively noisy measurements of sensors with limited capability. The goal is to identify multiple materials that are embedded in a particular pixel and thus, provide for more effective image fusion. However, without prior knowledge of a scene, it is difficult, if not impossible, to identify and classify the number of separate sources in a scene of interest. To address this issue, a method of using independent component analysis (lCA) techniques to solve the unsupervised blind source separation (BSS) problem on hyperspectral data was implemented. This general methodology offers the capacity to determine what endmembers are present in a scene and their levels of existence within individual pixels. The FastICA algorithm was used in conjunction with A VIRIS hyperspectral data to demonstrate this method.