分布式X光源由若干碳纳米管阴极X射线源集成而成，是一种新型的多光束源，和传统射线源相比，它具有结构紧凑、脉冲响应、可编程出束等优点。基于分布式X光源的动态CT设备依靠射线源焦点依次快速出束完成扫描，不需要机械运动，因而时间分辨率高。由于工作环境恶劣，固体发动机喉衬等部位容易发生烧蚀现象，造成严重后果。动态CT可以在极短时间内完成扫描，在这段时间内，物体处于准静态状态，能够对多相流、固体发动机燃烧过程等进行过程成像。本论文从固体发动机燃烧过程动态成像检测的需求出发，研究基于分布式X光源的动态CT成像技术，主要研究内容包括：论文首先在充分调研现有动态CT技术基础上，研究了基于分布式X光源的动态CT成像系统技术方案，并依据工业CT设计原理，通过数值仿真模拟，计算CT成像系统各项技术参数，完成分布式X光源动态CT系统设计。论文针对分布式X光源动态CT系统中投影数据缺失情况，重点研究了迭代图像重建算法，引入基于压缩感知的总变分最小化正则约束，采用SIRT-TV、OSEM-TV等算法获得了良好的重建效果；同时为了加快动态CT图像重建速度，研究采用CUDA来加速迭代重建计算过程。论文通过搭建动态CT实验平台，借助现有条件开展了系列实验来验证动态CT设计方案以及图像重建算法，并针对重建图像噪声问题，研究了双边滤波去噪以及小波阈值去噪的图像后处理方法。本课题为固体发动机燃烧等动态过程研究工作提供了一种新型动态CT成像方法，具有广泛的应用前景。
The distributed X-ray source is a novel multi-beam source which consists of several carbon nanotube cathode source units. It provides the advantages of compact structure, impulse response, programmable control over conventional X-ray tube. The dynamic CT based on distributed X-ray source has a high temporal resolution for no mechanical movement. Scanning process can be completed in a short time relying on focus spot firing in turn. Solid engine is widely used in many fields, its throat linings and other parts are prone to be ablated due to terrible operation environment. Scanning process can be completed in a very short time with dynamic CT. During this time, the object is in a quasi-static state. It can be used to image the multi-phase flow, solid engine combustion process. In this paper, we research the dynamic CT technology based on distributed X-ray source for the requirement of dynamic inspection in the combustion process of solid engine. The main research contents include:Based on the full investigation of the existing dynamic CT, we studied the distributed X-ray source dynamic CT technology. System design is completed based on the industrial CT design principles. We calculate the technical parameters of the CT system with the help of numerical simulation platform. We focus on iterative reconstruction algorithms to solve the incomplete projection data problem, total variation minimization regularity constraint based on compressive sensing is introduced. We get a good reconstruction result by using SIRT-TV, OSEM-TV and other algorithms. In order to speed up the image reconstruction, CUDA is used to accelerate the iterative reconstruction calculation process. We build a distribute X-ray source dynamic CT prototype, and a series of experiments is conducted. Design scheme and image reconstruction algorithms are verified. Bilateral image denoising and wavelet threshold denoising image post-processing methods are adopted. This project provides a new dynamic CT imaging method to inspect dynamic process of solid engine combustion. It has a wide range of application prospects.
