Image de-noising via spectral distribution similarity analysis for ultrasonic non-destructive evaluation
Image de-noising via spectral distribution similarity analysis for ultrasonic non-destructive evaluation
复制标题
通过光谱分布相似性分析进行图像去噪以进行超声无损评估
DOI：
--
复制
发表时间：
2014
期刊：
影响因子：
--
通讯作者：
A. Gachagan
中科院分区：
文献类型：
--
作者：
Bo Xiao;Minghui Li;Rui Gongzhang;R. O’Leary;A. Gachagan
在许多行业中，对表现出异质和散射微观结构的粗粒材料中的缺陷进行超声检测和表征特别重要，但仍然具有挑战性。文献中的大多数基于谱的去噪方法对材料特性敏感，这需要一个麻烦的参数优化过程，从而阻碍了它们在超声图像处理中的应用。为了提高图像中缺陷的可见性，提出了一种基于谱分布相似性分析（SDSA）的杂波抑制方法。该方法等距分段所有的时间序列数据在一个数据集获得的全矩阵捕获技术，然后普查的光谱分布的全局段和局部段的每一个聚焦点的总聚焦方法的图像。通过测量两个光谱分布之间的相似性计算的系数揭示了合理的缺陷指示的可能性。
Ultrasonic detection and characterization of flaws in coarse-grained materials exhibiting heterogeneous and scattering microstructure is of particular importance across many industries, but remains challenging. Most spectral based denoising methods in the literature are sensitive to material properties, which necessitate a troublesome parameter optimization process and consequently impede their application into ultrasonic image processing. In order to improve flaw visibility in an image, we propose a novel and robust clutter suppression method through spectral distribution similarity analysis (SDSA). This method isometrically segments all the time-series data in a dataset acquired by the Full-Matrix-Capture technique and then censuses the spectral distribution of global segments and of local segments for every focusing point in the Total-Focusing- Method image. The coefficient computed by measuring the similarity between the two spectral distributions reveals the possibility of a legitimate flaw indicatio...