基于ATV与联合双边滤波的NSCT声呐图像去噪OACHSSCD

To tackle the problem of edge detail loss caused by speckle noise in sonar imaging,a multi-scale fusion-based composite denoising framework was proposed.The method establishes a dual-domain collaborative processing mecha-nism based on the non-subsampled contourlet transform(NSCT).Firstly,by introducing dual parameters—average gradi-ent and edge intensity—dynamically adjusted penalty factors for the anisotropic total variation(ATV)model were con-structed to apply to high-frequency sub-bands.Subsequently,the structural information obtained from ATV processing acted as guidance maps to drive a joint bilateral filter for edge-aware denoising,which simultaneously suppresses multipli-cative noise and builds a barrier for detail preservation.For low-frequency coefficients,bilateral filtering was employed for smoothing adjustment to effectively control residual noise.Finally,the processed high-frequency and low-frequency co-efficients were reconstructed through the inverse NSCT to obtain image denoising.The experimental results showed that,compared with wavelet transform-based denoising,traditional Split Bregman TV denoising,DWT threshold denoising,and NSST domain denoising,the proposed method more effectively reduces speckle noise,retains more image edges,and im-proves the visual quality of images.