Research on SAR image denoising method based on feature extraction
Shaoming Wei,Xin Ma,Fangrui Qu,Jun Wang,Tian Liang,Dehong Chen
DOI: https://doi.org/10.1049/ell2.13180
2024-04-27
Electronics Letters
Abstract:This paper proposes an SAR image reconstruction model that addresses speckle noise through image denoising and feature extraction. The model calculates singularity indices and extracts informative features to reconstruct a denoised image, achieving excellent filtering effects while preserving texture information. Evaluation results demonstrate its effectiveness in reducing background clutter and enhancing target‐background contrast. Due to the excellent characteristics of Synthetic Aperture Radar (SAR), it has become an important technical means for land and sea target detection applications worldwide. However, the SAR image imaging process is often accompanied by speckle noise, which affects the accuracy of target detection. For this problem, this paper proposes an SAR image reconstruction model based on image denoising and feature extraction. Given a noisy SAR image, at each pixel of the image, the value of the singularity index of a metric defined by the signal gradient norm density is calculated. Then through a hierarchical decomposition of these singular indices, a set of pixels containing the most informative feature set of the image is extracted, and a denoised image is reconstructed using the extracted features. In comparison to traditional filtering and denoising techniques such as non‐local filtering, this algorithm achieves remarkable local filtering and denoising effects while preserving the image's texture information. The denoising performance of the proposed method is evaluated through a series of quantitative indicators and visual comparisons. The results demonstrate that this method effectively reduces background clutter, resulting in a higher contrast between the target and the background.
engineering, electrical & electronic