Research Of Flame 3D Reconstruction Based On Computed Tomography And Compressive Sensing

In recent years,the air pollution caused by automobile exhaust has become more and more serious.It has become one of the important causes for fog and haze.Energy saving and emission reduction of internal combustion engine has attracted global attention.As an important representation of combustion process in the cylinder of internal combustion engine,combustion flame is of great significance to study the mechanism of flame combustion,reduce the exhaust gas combustion and improve the performance of internal combustion engine.The development of modern optical methods and computer technology make it possible to utilize the optical visualization technology to observe the combustion process of internal combustion engine directly.The study on 3D shape of flame is beneficial to understand the characteristics and laws of combustion in cylinder.So it is necessary to reconstruct the 3D shape of the flame to providing sufficient information for the quantitative analysis of the flame.In the 3D image reconstruction algorithm,the Computed Tomography and the Compressed Sensing are two classical algorithms.The key issue of this paper is how to reconstruct flame image effectively with CT algorithm and CS algorithm.In this paper,we analyze the special properties of flame and the principle of traditional flame 3D reconstruction based on CT algorithm and CS algorithm.According to the gradient sparsity characteristics in the Total Variation(TV)domain and radial diffusion characteristics of engine combustion flame,a flame 3D reconstruction algorithm with higher stability and reconstruction accuracy is proposed.It is the Algebraic Reconstruction Technique(ART)reconstruction algorithm based on radial TV(ART-RTV algorithm).And based on this algorithm,the flame 3D reconstruction visualization is carried out.Numerical simulation results show that the new proposed ART-RTV algorithm can reconstruct flame 3D images more stably and has a better robustness than the traditional ART algorithms especially in limited-angle situation.