A varying-energy CT reconstruction method for improving reconstruction of metal workpieces with highly variable thickness

In industrial computed tomography (CT) applications, it is very challenging to reconstruct high-density metal workpieces with highly varying thickness to obtain all the structural information by single-energy CT imaging due to the existence of underexposed or overexposed projections. However, by using varying x-ray energy to scan the workpiece, the acquired projections share structural similarity, which can complement each other. In this work, we present a varying-energy imaging method to reconstruct a single-material workpiece of highly varying thickness. Considering that the practical acquired projections are poly-energetic, causing the beam-hardening problem, the projections acquired at each scanning energy are firstly decomposed to obtain the approximate mono-energetic projections. Then the decomposed projections are fused by a projection fusion algorithm based on a scaling transformation and residual smoothing model, followed by a conventional reconstruction method. The practical experiment results indicate that the proposed algorithm can effectively improve the quality of the reconstructed image compared to the single-energy CT reconstruction method, by fusing the valid information of multi-energy projections.