High resolution remote sensing image denoising based on Curvelet-Wavelet transform

Abstract

A novel donoising method in multi-domain based on cartoon and texture decomposition model was proposed to remove the additional noise from remote sensing image.The algorithm considered that Curvelet transform and Wavelet transform have different sparse representation features for different image parts.An undecimated version of orthogonal Wavelet and a wrapping-based Curvelet transform were used.Curvelet coefficients and Wavelet coefficients were constructed by using image decomposition model,and Gaussian scale mixtures(GSM)model was combined to remove image noise.Experimental results show that the method can effectively preserve image details and edges as well as remove pseudo-Gibbs phenomena,and the peak signal-to-noise ratio evidently increases.