Speckle suppression of synthetic aperture radar image with empirical mode decomposition and principal component analysis based on noise energy estimation
Xiangli Wang, Li Layuan · Journal of Applied Remote Sensing · 2015
We present a method to reduce speckle noise in synthetic aperture radar (SAR) images based on empirical mode decomposition (EMD) and principal component analysis (PCA). First, the logarithmic SAR image is decomposed by EMD. Using the statistical characteristics of logarithmic speckle noise and energy distribution model of EMD-decomposed white noise, we estimate the energy magnitude of noise in each level intrinsic model function. Second, PCA was adapted to process the intrinsic mode function of each level. After the intrinsic mode function was decomposed by PCA, a part of the principal components is abandoned according to the proportion of noise energy in the intrinsic mode function for further removing the noise of the intrinsic mode function, and the intrinsic mode function is reconstructed by the remaining principal components. Finally, the denoised SAR image is obtained by accumulating all the processed intrinsic mode functions. Experimental results show that the proposed algorithm demonstrates a higher performance than traditional EMD algorithms in edge retention and speckle removal.