Development of Beam Hardening Correction Algorithms for Industrial Computed Tomography

This thesis is concerned with the use of X-ray computed tomography (CT) for industrial inspections and dimensional measurements. Beam hardening is a phenomenon that is encountered in the reconstructed CT images. It is known to deteriorate CT image quality. A number of beam hardening correction (BHC) algorithms have been developed over the past decades. However, most of them were developed in medical CT and there is a lack of research on their effectiveness on industrial CT. Starting with a thorough comparison of the state-of-the-art BHC algorithms in industrial CT, the author has gradually widened the scope of the research to a rigorous and comprehensive study on BHC in industrial CT. The main contributions of this thesis are summarised as follows: 1. An in-depth BHC algorithm comparison for industrial CT Since this study takes various factors of the algorithm performance into account, it provides insights into the advantages and disadvantages of each algorithm. 2. A benchmark of the dependence of the BHC methods on the spectral information Most of the BHC methods rely on prior knowledge of the X-ray spectra, which means the performance of them may be hindered if the spectral information is not accurate. This work addresses the need for an investigation of the spectral dependence of the BHC methods. 3. An X-ray spectral estimation framework and the analysis of various uncertainty sources X-ray spectral estimation using transmission measurements has great potential as the acquired spectral information can be directly utilised by a BHC algorithm. Nevertheless, the measurement process can be error-prone. This work analysed the various potential error sources of the proposed method and their influences on the measurement accuracy. As scattering is a main error source of spectral estimation, an innovative spectral estimation algorithm is further introduced that can yield accurate spectral estimation without the need for prior scattering correction. 4. Two new beam hardening correction (BHC) algorithms for multi-material industrial CT Based on the observations of the BHC algorithm comparison studies, the Joseph and Spital (JS) segmentation method seems to have great potential for its high performance, robustness and speed. However, it is limited by the computer memory constraints in its application in multi-material objects. The first proposed algorithm employs a statistical learning strategy to reduce the memory consumption of the JS method. The second proposed algorithm presents a fully automated BHC workflow with the aid of neural network (NN). 5. A pipeline for optimised BHC results with the proposed method This work analyses the reconstructed image quality improvement with the current strategies. It involves data acquisition of a multi-material phantom and spectral measurement based on the Nikon XTH 450 system. The phantom dataset is acquired with narrow beam collimation to minimise scattering. By using the measured spectra as a priori information for the BHC algorithm, the remaining artefacts in the reconstruction image can be minimised.

Jaar van publicatie:2019

Toegankelijkheid:Open