Satellite Image Denoising via Adjustment of the Quaternionic Wavelet Coefficients

This paper describes an approach to images denoising using relationship between the quaternionic wavelet coefficients for use in remote sensing applications. These images have several spectral components and strong geometric information. The quaternionic wavelet transform gives a very good separation of the coefficients in terms of magnitude and 3angles phase and generalize better the concept of analytic signal to the image. Denoising is based the spatial- spectral mutual neighborhood where quaternionic magnitude coefficients are thresholded in local window. In addition to magnitude, the phase variations in the same scale are exploited in a global schema. Our method is applied to satellite images representing regions of Algeria. Results indicate that the performances have increased in noise suppression and edge preservation compared with the wavelet methods that do not use the phase or multichannel information.