共轭梯度法在图像重建中的应用
Conjugate Gradient Applied to Image Reconstruction
-
摘要: 图像重建常常被转化为解非线性无约束极值问题,通过范数极小化推导出共轭梯度法的一般算法。通过对模拟数据和实际工件断层扫描数据进行图像重建,估计了算法的有效性,结果表明,与最速下降法相比,此算法更适用于不完全投影数据的图像重建,在保证重建图像拟合度的同时,大大提高了重建速度。Abstract: Image reconstruction is often converted to the nonrestraint extremum of the nonlinear equations.We have deduced a conjugate gradient algorithm from the extremum of norm.The experimental results of reconstruction from simulated projections and real tomography projections have successfully shown that conjugate gradient guarantees the fitness to original image,speeds the iteration time,and can be specially used for reconstruction for incomplete projections.