超声散射CT信号的特征提取

Ultrasonic scattering computed tomography signals feature extraction

超声检测技术中,缺陷的定性分类这一技术难题至今尚未得以彻底解决。研究了希尔伯特-黄变换(HHT)和小波包变换(WPT)在超声散射CT特征提取中的应用。以荷兰试块作为测试对象,针对3个不同的检测面,采用HHT和WPT提取能量特征,采用基于距离的类别可分性判据对特征提取结果进行评价,起了定量衡量尺度的作用。实验结果表明:对于本实验中信号频率比较单一的超声散射CT信号,WPT提取的能量特征比HHT提取的能量特征更为有效。

The problem of flaw classification in ultrasonic testing has not been solved so far.The applications of the Hilbert-Huang Transform（HHT） and Wavelet Packet Transform（WPT） in ultrasonic scattering CT feature extraction were studied.Taking the test blocks for example,the energy features were extracted by HHT and WPT of three different inspection face,the characteristic values of measured echo signals are extracted evaluated by the separability criterion,and play a role of being a quantification measure.It is concluded that WPT is more valid than HHT in the fault feature extraction,especially under the condition that the ultrasound scattering CT signal frequency is single.