|
Noise Destriping in Hyperspectral Imagery in Frequency Domain by Combination of Wavelet and Fourier Transform
|
Y. Rezaei *   |
|
|
|
Abstract: (6737 Views) |
| Noise removing and radiometric correction is one of the challenging in Hyperspectral image processing. The one of these errors is striping which is presented in most of the remote sensing imagery. The destriping methods include statistical and filtering approaches. In the most of these algorithms, the structural information also removed after destriping. The presented method is combined wavelet-FFT filter in order to remove stripe artifact problem. In the first step, the original image is wavelet decomposed and subsequently, the bands containing the stripe information (vertical detail) are FFT transformed to remove the stripe errors. The visual assessments, as well as quantitative estimation of energy loss of the result show the capabilities and the performance of the purposed method in order to destriping. Also the result shows all structural features, which are different from stripes are optimally preserved and despite the statistical methods, the purposed algorithm doesn’t need the neighborhood information. |
|
| Keywords: Hyperspectral imagery, Stripe noise, Wavelet, FFT |
|
|
Full-Text [PDF 974 kb]
(1992 Downloads)
|
Type of Study: Research |
Subject:
Photo&RS
Received: 2014/07/21
|
|
|
|
|
|
|
| Send email to the article author |
|
|
|
| Add your comments about this article |
|
|
|
