四种系统矩阵的建模方法在ASD-POCS算法中的比较

Comparison of four system modeling methods in ASD-POCS

针对在图像重建算法中,系统矩阵的建模方法对重建效果有一定影响的问题,设计并实现了四种系统矩阵建模方法,即像素驱动、距离驱动、改进的距离驱动以及射线驱动,并且将它们与ASD-POCS算法相结合,比较了它们的建模速度、建模精度以及收敛速度,同时比较了它们在有噪声的情况下对噪声的敏感程度,以及不同系统矩阵建模方法在ASD-POCS中的重建效果。实验表明,距离驱动的建模速度最快,而改进的距离驱动的建模精度最高,射线驱动在ASD-POCS算法中的收敛速度最快。

In image reconstruction algorithms, there are several difference exist by using kinds of system matrix modeling me-thods.In order to solve this problem, this paper designed and finished these methods：pixel driven, distance driven, improved distance driven and ray driven.Then it combined them with ASD-POCS and compared their speed of modeling, accuracy and their convergence rate in ASD-POCS algorithm.It also compared their sensitive degree by adding noise in projection data.Experiments prove that, distance driven is faster than others in system modeling, improved distance driven is more accurate and the ray driven has the fastest convergence speed.