Optimal band selection strategies for hyperspectral data sets

Pacific Northwest National Laboratory Remote Sensing group, in cooperation with the West Virginia University, have been developing, testing and applying hyperspectral data sets ranging from early AVRIS data sets to HYDICE systems to characterize subtle landcover features. For example, using high resolution HYDICE data it is possible to discern subtle spectral differences among vegetation types that differentiate noxious weeds. This differentiation was not possible with low resolution multispectral data sets. However, the large amounts of data that must be processed present significant problems to the end user. These include increased operational complexity and cost, computational burdens, data transmission bandwidth limitations, and loss of capabilities that might be offered through system trade-offs. In order to help overcome these problems a number of band selection strategies were developed and tested that significantly reduced the requirement to process and analyze redundant data. The authors have also investigated extensions of other workers' band selection methodologies that have proven useful. In general the application of these methods suggests that the number of bands needed to solve a given problem can often be reduced to 10 or less. In many cases this reduction in bands can remove significant barriers to the application of hyperspectral data to a large number of environmental problem sets.