International Society for Industrial Process Tomography

Multi-frequency Ultrasound Tomography based on Modified Regularization Method and Wavelet Fusion

Wenxiu Hou, Chao Tan^*, Yong Bao, Feng Dong

Tianjin Key Laboratory of Process Measurement and Control

School of Electrical and Information Engineering, Tianjin University, Tianjin, China

^*Email: tanchao@tju.edu.cn

ABSTRACT

Ultrasound tomography (UT) is an emerging technique for the industrial process detection, different media in the measurement zone can be identified by the acoustic characteristics at ultrasound excitation. This paper presents an investigation of multi-frequency UT and the novel reconstruction algorithm. A multi-frequency UT system has been set up for the investigation which can realize the 32 channels ultrasound signal excitation and reception. A linear frequency modulation signal is chosen for the multi-frequency ultrasound excitation and the multi-frequency ultrasound receiving signals are demodulated and further transform to the single frequency information. The proposed reconstruction algorithm combines the modified regularization method and wavelet fusion method. The common Tikhonov regularization algorithm is modified by replacing the identity matrix to the difference matrix using second-order Laplacian operator. The modified regularization method can effectively reduce the artefact for the inclusion and sharpen the boundary compared to the other methods. The wavelet fusion is adopted for the single frequency images fusion as the receiving signals at different frequency excitation have different ultrasound responses depicted at the different frequency reconstruction images. The experimental results show that the multi-frequency UT with the fusion method using the bior3.7 wavelet function can obtain the better reconstruction quality than the single frequency reconstruction image.

Keywords: Ultrasound tomography; Multi-frequency; Liquid-solid two-phase media; Regularization; Wavelet fusion

Industrial Application: Liquid and solid