一种改进的变分正则化CT重建算法

An improved variational regularization CT reconstruction algorithm

针对常规ART算法重建图像像素值不稳定,收敛速度慢的缺点,从正则化方法和求解不适定方程问题的数学角度入手,提出了能应用在CT重建中的一种改进的自适应变分正则法.该算法首先对投影数据进行ART重建,然后引入重建图像的梯度作为先验信息,并让梯度自适应下降,来改善重建结果.仿真实验表明,本文算法不仅很好地重建了完整图像的结构信息,还使重建图像更加平滑,收敛速度更快.

Considering its slow convergence and unsmooth reconstruction image for conventional ART algorithm, proposed an adaptive variational regular algorithm from the mathematical point of regularization method and solving ill-posed equations, which can be used in CT reconstruction. To improve the reconstruction results, firstly, using ART algorithm reconstruct an image, then making gradient as a priori information and drop adaptively. Simulation experiments showed that the proposed algorithm can not only reconstruct complete structure information, but also make the image of reconstruction smoother and the time fewer.