基于VMD方法的混凝土缺陷超声成像噪声处理研究

Research on ultrasonic imaging noise processing of concrete Defects based on VMD method

摘要:

时间反演多信号分类(TR-MUSIC)算法进行混凝土结构缺陷超声成像,能够在激励超声频率不改变,保证探测深度的情况下,打破衍射限制,实现超分辨率成像；但此方法的缺点是极易受到噪声的干扰,在成像划分信号子空间与噪声子空间过程,产生附加奇异值,影响超声成像的精度和准确性；为进一步提高TR-MUSIC超声成像的质量和稳定性,抑制噪声对成像的干扰,提出一种基于变模态分解(VMD)的噪声处理方式,该方法首先对VMD分解后的变模态分量采用能量熵和重心熵结合的方式确定信号频谱形态发散和能量分布不均的噪声分量,对噪声分量采用小波阈值降噪方法处理,然后将降噪后的噪声分量和其他分量进行Savitzky-Golay filtering(最小二乘平滑滤波)并重构；改善后的信号进行反演成像,提高了信号子空间和噪声子空间对应奇异值大小的区分度,再通过设定阈值方式进行划分,提高了成像分辨率,抑制了由噪声引起的伪像问题。

Abstract:

Time inversion multi-signal classification (TR-MUSIC) algorithm for ultrasonic imaging of defects in concrete structures can break the diffraction limit and realize super resolution imaging under the condition that the excitation frequency does not change and the detection distance is guaranteed； However, the disadvantage of this method is that it is easy to be disturbed by noise, and additional singular values are generated in the process of imaging dividing signal subspace and noise subspace, which affects the accuracy and accuracy of ultrasonic imaging； In order to further improve the quality and stability of TR-MUSIC algorithm in ultrasonic imaging and suppress the interference of noise on imaging, a noise processing method based on variable mode decomposition (VMD) was proposed；The variable mode component after VMD decomposition was determined by combining energy entropy and barycentric entropy to determine the noise component of signal spectrum divergence and uneven energy distribution. The noise components were processed by wavelet threshold denoising method, and the noise components and other components were processed by Savitzky-Golay filtering and reconstructed. Inversion imaging of the improved signal improves the distinction of the corresponding singular values of the signal subspace and noise subspace. Then, the imaging resolution is improved and the artifact problem caused by noise is suppressed by setting thresholds.