【摘要】 Radon变换反演分辨率是其在地震资料处理中应用的关键因素.常规高分辨率Radon变换反演模型采用L₁范数实现稀疏正则化，其以Radon系数振幅总体大小衡量稀疏性，未能充分考虑到Radon系数振幅相对强弱，以此构建的反演方法旨在通过衰减弱振幅来提高稀疏性，对Radon系数分辨率提高效果有限.本文提出一种以信号振幅相对大小衡量Radon系数稀疏性的尺度不变稀疏度量（Scale-Invariant Sparsity Measure, SSM）函数，并以L₁/L₂范数作为其光滑近似.针对L₁/L₂正则化的非线性，改进交替方向乘子法（Alternating Direction Method of Multipliers,ADMM）对其求解.合成数据与实际地震资料处理结果表明，该方法能够增强一次波与多次波在Radon域的分辨率，降低Radon变换的能量扩散，提高多次波压制效果；同时该方法能够较好地压制随机噪声，方法具有一定的稳健性.更多还原

【Abstract】 The resolution of Radon transform inversion is a crucial factor in its application in seismic data processing. The L₁-norm is used as the sparse regularization term in the modeling of conventional highresolution Radon transform inversion method. The sparsity is measured by the overall size of the amplitude of the Radon coefficient. This fails to fully consider the relative strength of the amplitude of the Radon coefficients at each frequency. The inversion method constructed by this regularization method improves the sparsity by suppressing the amplitude, and the performance of improving the resolution of Radon coefficients is limited. In this paper, the Scale-Invariant Sparsity Measure（SSM） function is proposed to measure Radon coefficients by the relative magnitude of signal amplitude, and the L₁/L₂-norm is used as its smooth approximation. By using L₁/L₂-norm regularization to constrain sparsity, a high-resolution Radon transform inversion model that can effectively improve the focusing performance of coefficients is constructed, and the Alternating Direction Multiplier Method（ADMM） is used to solve it. The processing results of synthetic data and field seismic data show that the proposed method can enhance the resolution of primary and multiple events in the Radon domain,weaken the energy diffusion in the Radon domain, has good noise resistance, and can improve the multiple attenuation performance.更多还原