多色X射线同轴相衬CT射束硬化校正和图像复原算法的研究

In polychromatic x-ray in-line phase-contrast computed tomography, the effects of beam hardening and degradation of projection image are the main error sources of reconstruction and have seriously hindered practical application of the technique. The item will study on the correction of beam hardening and image restoration in polychromatic x-ray in-line phase-contrast computed tomography using a computed tomography system based on microfocus x-ray sources. First, we will research on the way beam hardening influence the reconstruction. Based on the results, we will develop our beam-hardening correction methods for polychromatic x-ray in-line phase-contrast computed tomography of single-component and multi-component materials. Then a fast algorithm of image restoration with ability of edge-preserving and inhibition of noise-amplification and ringing is developed for polychromatic x-ray in-line phase contrast computed tomography. The method combined the wavelet transform and adaptively regularized image restoration. The beam-hardening correction and image restoration methods are further accelerated with CUDA technique. The analysis and methods presented in this item poeeess important theoretical significance and application value.

在多色X射线轴相衬CT中，射束硬化和投影图像质量退化引起重构准确度的下降和伪影的出现，严重影响了多色X射线相衬CT技术的实际应用。本项目以微聚焦X射线同轴相衬CT系统为研究平台，研究多色X射线同轴相衬CT射束硬化校正和图像复原。研究射束硬化对多色X射线同轴相衬CT的影响机制，并在此基础上探索单材质和多成分样品的多色X相衬CT射束硬化校正算法。针对同轴相衬CT，开发一种基于小波变换的自适应图像复原算法，该方法是一种能有效抑制边缘振荡、边缘模糊和噪声放大的快速图像复原算法。研究中还将利用统一计算设备架构技术，以进一步提高射束硬化校正算法和图像复原的计算速度。这些方法的研究和建立具有重要的理论意义和应用价值。