Baselice Fabio
Dipartimento di Ingegneria, Università degli Studi di Napoli Parthenope, Naples, Italy.
Ultrasound Med Biol. 2017 Sep;43(9):2065-2078. doi: 10.1016/j.ultrasmedbio.2017.05.006. Epub 2017 Jun 23.
Ultrasound images are affected by the speckle phenomenon, a multiplicative noise that degrades image quality. Several methods for denoising have been proposed in recent years, based on different approaches. The so-called non-local mean is considered the state-of-the-art method; the idea is to find similar patches across the image and exploit them to regularize the image. The method proposed here is in the non-local family, although instead of partitioning the target image in patches, it works pixelwise. The similarity between pixels is evaluated by analyzing their statistical behavior, in particular, by measuring the Kolmogorov-Smirnov distance between their distributions. To make this possible, a stack of acquired images is required. The proposed method has been tested on both simulated and real data sets and compared with other widely adopted techniques. Performance is interesting, with quality parameters and visual inspection confirming such findings.
超声图像会受到斑点现象的影响，这是一种会降低图像质量的乘性噪声。近年来，基于不同方法提出了几种去噪方法。所谓的非局部均值被认为是最先进的方法；其思路是在图像中找到相似的图像块，并利用它们来对图像进行正则化。这里提出的方法属于非局部方法族，不过它不是将目标图像划分为图像块，而是逐像素进行处理。通过分析像素的统计行为，特别是通过测量它们分布之间的柯尔莫哥洛夫-斯米尔诺夫距离来评估像素之间的相似性。为此，需要一组采集到的图像。所提出的方法已在模拟数据集和真实数据集上进行了测试，并与其他广泛采用的技术进行了比较。性能表现令人关注，质量参数和视觉检查都证实了这些结果。
