A Novel SAR Image Despeckling Method Based on Local Filter With Nonlocal Preprocessing
Chao Wang,Baolong Guo,Fangliang He
DOI: https://doi.org/10.1109/jstars.2023.3258424
IF: 4.715
2023-04-01
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Abstract:Owing to the characteristics of long distance and strong penetration, a synthetic aperture radar (SAR) imaging system could provide ground information with high resolution under a poor climate environment. Nevertheless, speckle is still a common interference of the output that deteriorates the content of SAR images and further affects the recognition of real objects. In this article, a new speckle suppression method is proposed from the perspective of exploring nonlocal and local SAR image features. Considering the statistical distribution of SAR images, a novel local filter termed SAR-orientated guided bilateral filter is proposed to characterize the range and spatial similarity of SAR images. Meanwhile, an optimized nonlocal filter based on the weight Schatten- norm is introduced to characterize the nonlocal self-similarity of SAR images by a low-rank model. As a preprocessing step, it yields nonlocal filtering features as the guidance image of the proposed SAR-oriented guided bilateral filter. By incorporating the nonlocal filtering feature into the local filter, the structured method could achieve desirable despeckling results. Extensive experiments on real SAR images demonstrate that the proposed method outperforms several state-of-the-art methods in terms of both visual satisfaction and quantitative metrics.
imaging science & photographic technology,remote sensing,engineering, electrical & electronic,geography, physical