XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE
Image Reconstruction Based on Regularized
Weighted Least Square Framework for Low-
Contrast Ultrasonic Tomography
Hao Liu, Chao Tan, Feng Dong
Tianjin Key Laboratory of Process Measurement and Control
School of Electrical and Information Engineering, Tianjin
University
Tianjin, China
E-mail: fdong@tju.edu.cn
Shu Zhao
Institute of Biomedical Engineering
Chinese Academy of Medical Sciences and Peking Union
Medical College
Tianjin, China
Email: zhaos@bme.cams.cn
Abstract—Two-phase flow is a flow regime that commonly
existed in many applications of process industry, on which the
technique of Ultrasonic tomography could provide reliable
measurement with benefits of non-invasion, non-radiation and
low-cost. In order to provide considerable visualization of low
acoustic impedance contrast configurations in two-phase flow,
a novel reconstruction approach using continuous-wave
excitation is proposed and presented in detail. With the
measurement strategy introduced and the reconstruction
model formulized, the system matrix assemble method is
modified and the regularized weighted least square framework
combining TV regularization and SART reconstruction is
derived. Both numerical simulation and experimental
investigation are conducted to validate the feasibility of
continuous-wave excitation and the improve of proposed
reconstruction strategy. In the imaging results, the distribution
of oil-water two-phase flow is effectively reconstructed with
relative error (RE) reduced 27.85% and correlation coefficient
improved 26.94%. Both the system matrix assembles method
modification and the proposed reconstruction approach
provides significant solution for the further development of
related ultrasonic tomography technique.
Keywords—Ultrasonic tomography; low impedance contrast;
continuous-wave excitation; system matrix assemble; regularized
least square framework
I. INTRODUCTION
Process Tomography is a rapidly evolving measurement
technique for multiphase flow imaging and is greatly
required in production process and industrial control due to
its capability to provide spatial distribution of different
phases in an opaque domain. Among all the modalities
existed, Ultrasonic Tomography (UT) is a non-intrusive and
non-invasive sensing technique because ultrasonic wave
owns the capability of penetrating through vessels and fluids,
leading to the advantages of low costs and relatively simpler
sensing structure. The application of Ultrasound Tomography
also has negligible safety issues compared to ionizing
radiation method like X-ray CT or Gamma-ray CT [1].
Based on the rich information contained in ultrasound
waves and its benefits in industrial measurement, there has
been significant interest in the usage of ultrasonic sensors for
multiphase flow tomographic imaging. For instance, Xu
presents an ultrasonic tomography system to identify the flow
regime and void fraction of gas/liquid two-phase flow [2].
Later, the design of ultrasonic tomography system and
supportive reconstruction strategy is proposed to image the
gas/liquid two-phase flow distribution while the feasibly of
dual-modality combining transmission and reflection is
discussed [3]. Further improvement of Ultrasonic
Tomography is being conducted, from image reconstruction
using offline approaches, to online system and different
ultrasonic sensing approaches, such as multi-wave
measurement, fan beam transducers, and simultaneous-
excitation [4]. There is continuous growth of interest
focusing on ultrasonic tomography for two-phase flow
imaging in recent years.
Although with such advantages, the application of
Ultrasonic Tomography still poses the challenge of low
projection number and reduced performance on low contrast
inclusions. As mechanical wave, ultrasonic wave needs time
to penetrate through the testing domain. The number of
transducers has to be restricted to meet the requirement of
real-time measurement, leading to a low resolution. On the
other hand, the scenario of low acoustic impedance contrast
is also widely existed in two-phase flow (like oil/water two-
phase flow) and its reconstruction relies on accurate
measurement of amplitude attenuation, which could be
hardly managed through conventional pulse-wave excitation.
There hasn’t been considerable progress in ultrasonic
tomography for low impedance contrast scenarios.
Based on above discussion and focusing on accurate
reconstruction of low impedance contrast distribution in
ultrasonic tomography, a novel ultrasonic tomography
approach using continuous wave excitation is proposed and
presented in detail. It aims at the visualization measurement
of transmission attenuation coefficient in two-phase flow
with low acoustic impedance contrast. With the basic
principle of sequential excitation introduced and “one-to-all”
measurement strategy discussed, the theoretical correlation
using continuous-wave excitation is introduced through a
formulized reconstruction model. To provide accurate
reconstruction, the design of distance-modified fan shape
system matrix is proposed and the application of regularized
weighted least square (RWLS) method combines the total
variation (TV) regularization and simultaneously algebraic
reconstruction technique (SART) adaptively. The RWLS
framework provides the possibility of combining
regularization method and algebraic reconstruction to solve
the under-determine situation and ill-conditioned problem at
the same time. Both simulation and experimental
investigation are conducted using the proposed UT approach
to reconstruct oil/water two phase configuration inside a
Plexiglas pipe, resulting in a preferable reconstruction result
with higher imaging accuracy and less artifacts.
The authors appreciate the support from the National Natural Science
Foundation of China (No. 61571321) and the Natural Science Foundation
of Tianjin (No. 17JCZDJC38400).