Wavelet De-Noising of Hyperspectral Data
Abstract
This paper presents a method for noise removal from hyperspectral data using wavelets. Normally, data collected in the field will contain several types of noise. One type is the small amount of noise that is superimposed along the signal. This noise will not degrade the characteristic shape of the signal, but will add a small systematic disturbance to it. This noise is usually caused by the physical instrument or field environment conditions. This noise, although small, can be troublesome in interpolating data. The natural growth in the understanding of wavelet applications now affords the capability to remove this noise without degrading the signal. A wavelet transform was applied to hyperspectral percent reflectance - wavelength data sets. The resulting power spectrum was filtered so that specific wavelet coefficients were removed. An inverse wavelet transform was then applied to this filtered spectrum to obtain a noise free data set.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1996
- Accession Number
- ADA355101
Entities
People
- E. Simental
- T. Evans