AGRI遥感图像中条带噪声的分析与去除

Analysis and Removal of Stripe Noise in AGRI Remote-Sensing Images

分析了多通道扫描成像辐射计(AGRI)条带噪声的主要来源,建立了条带噪声的图像退化模型,提出了一种基于直方图匹配与各向异性全变分正则化相结合的去条带(HMATV)方法。该方法首先使用直方图匹配抑制探测器像元间的非均匀性响应,接着利用各向异性全变分正则化模型去除剩余的条带噪声。使用定性和定量指标对各方法的处理结果进行评价,结果显示:与其他现有前沿的去条带方法相比,所提方法不仅获得了更优异的条带噪声去除效果,还有效保护了原始图像的细节信息。

In this paper, the primary sources of AGRI stripe noise are analyzed, and an image-degradation model for this noise is established. A method of stripe-noise removal based on histogram matching and anisotropic total-variation regularization is proposed. The method first implements histogram matching to suppress the nonuniformity response between detector pixels, and then implements the anisotropic total-variation-regularization model to remove the remaining stripe noise. Qualitative and quantitative indices are used to evaluate the processing results of various methods. The evaluation results show that, compared with the existing leading stripe-removal algorithms, the proposed method achieves a superior stripe-noise-removal effect while effectively protecting the details of the original image.