The invention discloses a remote sensing image denoising method. The method comprises the following steps of logarithm transformation, double-density dual-number complex wavelet decomposition, shrinkage coefficient calculation, layer-by-layer processing, double-density dual-number complex wavelet reconstruction, inverse-logarithm transformation and image output. Double-density double-tree complexwavelet transformation, the Bayesian theory and the quantum-derived idea are combined, and double-density double-tree complex wavelet transformation has translation invariance and the anti-aliasing characteristic, is more approximate to continuous wavelet transformation and has higher directivity, so that a mutual promotion effect is achieved after double-density double-tree complex wavelet transformation is combined with the quantum-derived method, the contradiction between spot removing smoothing and detail maintaining is better solved, the quality processed images is improved, an image speckle removing method is simple and effective, the image detail maintaining capability is high, a better visual effect is achieved, high-frequency signals and noise of the images are effectively distinguished, the self-adaptability of local filtering is enhanced, and image edge details are kept to a greater extent.