Speckle is an unavoidable noise-like phenomenon in Synthetic Aperture Radar (SAR) imaging. In order to remove speckle, many despeckling methods have been proposed during the past three decades, including spatial-based methods, transform domain-based methods, and non-local filtering methods. However, SAR images usually contain many different types of regions, including homogeneous and heterogeneous regions. Some filters could despeckle effectively in homogeneous regions but could not preserve structures in heterogeneous regions. Some filters preserve structures well but do not suppress speckle effectively. Following this theory, we design a combination of two state-of-the-art despeckling tools that can overcome their respective shortcomings. In order to select the best filter output for each area in the image, the clustering and Gray Level Co-Occurrence Matrices (GLCM) are used for image classification and weighting, respectively. Clustering and GLCM use the co-registered optical images of SAR images because their structure information is consistent, and the optical images are much cleaner than SAR images. The experimental results on synthetic and real-world SAR images show that our proposed method can provide a better objective performance index under a strong noise level. Subjective visual inspection demonstrates that the proposed method has great potential in preserving structural details and suppressing speckle noise.

No abstract

Coherent imaging systems, such as synthetic aperture radar (SAR), often suffer from granular speckle noise due to inherent defects, which can make interpretation challenging. Although numerous despeckling methods have been proposed in the past three decades, SAR image despeckling remains a challenging task. With the extensive use of non-local self-similarity, despeckling methods under the non-local framework have become increasingly mature. However, effectively utilizing patch similarities remains a key problem in SAR image despeckling. This paper proposes a three-dimensional (3D) SAR image despeckling method based on searching for similar patches and applying the high-order singular value decomposition (HOSVD) theory to better utilize the high-dimensional information of similar patches. Specifically, the proposed method extends two-dimensional (2D) to 3D for SAR image despeckling using tensor patches. A new, non-local similar patch-searching measure criterion is used to classify the patches, and similar patches are stacked into 3D tensors. Lastly, the iterative adaptive weighted tensor cyclic approximation is used for SAR image despeckling based on the HOSVD method. Experimental results demonstrate that the proposed method not only effectively reduces speckle noise but also preserves fine details.