Lifting wavelet transform-based algorithm of SAR and optical images fusion

摘要

Present an image fusion algorithm based on Lifting Wavelet Transform to integrate disparate and complementary data between SAR and optical images. Lifting Wavelet Transform enjoys the advantage of saving memory and calculating time in contrast with Discrete Wavelet Transform. Also, a Gaussian smoothing filter is employed for reducing the speckle noises of SAR image before wavelet transform. In order to preserve the spatial and spectral resolution, different regional fused rules are proposed to fuse wavelet approximation and detail coefficients. In terms of details, the standard deviations are compared and the higher standard deviation samples are picked; to fuse approximation, the modulus maxima selection rule or weighted averaging rule is adopted through the Global Absolute Deviation of the Mean as a threshold. Finally, Fusion Factor and Fusion Symmetry are employed for evaluating the fused algorithm.