Ultrasonic CT Reconstruction Of Temperature Field Based On Iterative Method

Ultrasonic computed tomography (CT) is a new non-contact measurement method. It has become a hot topic in nondestructive measurement with the advantages of low cost, high efficiency and harmlessness to the human body. It has been applied extensively in industrial measurement, geological survey, medical imaging and other fields. In this work, iterative reconstruction methods of ultrasonic CT are studied, and the problem of temperature field reconstruction is solved. Ultrasonic CT methods are widely applied to temperature measurement, atmosphere monitoring and so on. So this topic has important theoretical significance and broad application value.As an important part of Ultrasonic CT, reconstruction algorithm ensures the accuracy of measurement. The basis of reconstruction algorithm is to calculate ultrasonic speed information using collected data. And the physical quantities under test are then gained from ultrasonic speed information. Reconstruction algorithms, especially iterative algorithms of tomographic reconstruction are discussed in this paper. And a lot of simulations are carried out to analyze the features and applicable scope of these methods.In this paper, the main work includes:(1) The concept and basic principle of ultrasonic CT is expounded in this paper. And the existing algorithms, including filter back projection, least squares method, singular value decomposition, Tikihonov regularization, algebra reconstruction technique, simultaneous iterative reconstruction technique, and maximum likelihood expectation maximization algorithm are studied. To test the algorithms, all the algorithms above are experimented in air temperature reconstruction. In the experiment, example data is simulated, the results are compared and discussed, and meanwhile the accuracy of each algorithm is presented, which provides useful information in practice.(2) Based on the non-minimization optimization theory, a new reconstruction algorithm is put forward. And this algorithm is introduced into the unimodal temperature reconstruction problem. The simulation result shows the superiority of the new algorithm. The main characteristics and scope of application of this algorithm are also analyzed.(3) Based on the Maximum-Likelihood Expectation-Maximization (ML-EM) iteration algorithm, a novel method of temperature distribution imaging is proposed. In this method, Gaussian distribution model is applied instead of the traditional Poisson distribution model to the ML-EM iteration algorithm. The simulation results prove the correctness and the validity of the proposed method. Especially in the occasion of large abnormal values in measurement data, the new algorithm is superior to the traditional one.(4) Based on the theory of Bregman iteration algorithm, a new correction method of ultrasonic CT is presented. In this method, an error equation of sparse matrix is constructed using the calculated speed values and the actual values. Then the equation is solved by Bregman iteration. And the reconstruction result is revised using the solutions of error equation. Numerical simulation results show that the correction method can eliminate most gross errors. Especially, the combination of correction method and the new ML-EM algorithm restrains errors on the boundaries, and shows strong abilities of noise removal and edge preservation.(5) The temperature and the wind velocity distributions are estimated simultaneously using Ultrasonic CT technology. And 3D simulations are also completed.