Proceedings of SPIE - the International Society for Optical Engineering
We reduced speckle noise in SAR imagery by retaining only those wavelet coefficients with significant third-order correlation coefficients. These coefficients were generated from the cross-correlation functions of the image and wavelet basis functions. Using this approach, we compared the results between directly applying our denoising method, and first preprocessing by taking the logarithm of an image. In our approach, we examined wavelet coefficients in an environment where the contribution from the second-order moment of the noise had been reduced.
Kozaitis, S.P., Ingun, A. Speckle denoising using wavelet transforms and higher-order statistics (2004) Proceedings of SPIE - The International Society for Optical Engineering, 5438, pp. 222-228.