Application Research Of CT Image Reconstruction Technology In The Rebuilding Of Scramjet Engine Flame

Computed Tomography(CT)technology is widely used in many fields because of its non-contact and accurate imaging advantages.Hypersonic vehicle is the strategic development direction of future military and civil spacecraft.As the key technology of hypersonic vehicles,scramjet engine is one of the hot focuses for all countries competing to develop.It has huge military value and economic value.The accurate reconstruction and detail depict of the scramjet engine flame based on CT image reconstruction technology can obtain the instantaneous distribution information of each component in the combustion process,and it has important reference value for the improvement of propellant combustion rate and engine structure.The goal of this research is to develop the CT image reconstruction algorithm for 3D flame reconstruction of scramjet engine based on the application of computed tomography technology in the field of combustion diagnosis of hypersonic vehicle.The CCD camera is usually used as a detector for the computed tomography of chemiluminescence system in the related fields.However,because of its high cost and the impossibility to build them up in large numbers,how to get high quality reconstructed images from sparse projections turns into the key for the development of flame reconstruction algorithm.The compressed sensing theory has proved that the original signal can be reconstructed accurately when the sampling frequency of the sparse signal is far lower than the Nyquist sampling law.This provides a new idea to explore the CT images reconstruction method under the condition of undersampling.Aiming at the incomplete characteristics of flame projection data,this paper proposes an iterative image reconstruction algorithm for flame reconstruction based on the theory of compressed sensing,referred to as CTF(computed tomography of flame)algorithm.This algorithm transforms the CT image reconstruction problem into a constrained optimization problem,and the iterative image reconstruction technique is used to solve it.Based on the 2D distance driven projection and backprojection technology,we design and implement the 2D slice-by-slice reconstruction algorithm.In addition,we extend it into 3D space and propose the direct 3D reconstruction algorithm.The projection process of 3D CTF algorithm is more consistent with the physical imaging process of actual system.It not only retains the excellent characteristics of 2D CTF algorithm,such as high computation accuracy and less resource occupation,but also relieves the coplanarrestriction of detector installation to a certain extent.The experimental results of simulation and actual projection data show that CTF algorithm can get high quality reconstructed images under sparse projection condition,and the reconstruction result of 3D CTF algorithm is more accurate and reliable than 2D slice-by-slice algorithm.Compared with the classical analytic image reconstruction algorithm,iterativealgorithm can perform much better in the case of low signal-to-noise ratio,incomplete projection data or uneven projection angle.However,the slow speed of iterative reconstruction is the biggest limit for it to be fully applied.In recent years,with the emergence of many new methods and theories in the field of signal and image processing,the development of massively parallel computing technology and the reduction of the cost of computer hardware,this limitation has been gradually broken.In this paper,based on the CUDA programming architecture,combined with the characteristics of the algorithm,the GPU parallel implementation of the CTF algorithm is realized.The test results show that the GPU parallel acceleration technology can greatly improve the efficiency of the image reconstruction of the CTF algorithm.In addition,two sets of CTF image reconstruction applications are developed in this paper to better support the practical application of the flame structure analysis of the scramjet engine.