一种基于Contourlet变换与分裂Bregman方法的CT图像重建算法

A CT Reconstruction Algorithm Based on Contourlet Transform and Split Bregman Method

摘要: 针对稀疏投影CT图像重建算法中梯度算子在图像稀疏表示方面的局限,本文将Contourlet变换与CT图像重构相结合,利用Contourlet变换包含图像中的方向信息,从而更有效表达图像这一优势,提出一种将Contourlet变换与最小化图像总变差(TV)方法相结合,进而借助分裂Bregman方法进行最优化求解的CT重建算法。实验结果表明,在投影数较少的条件下,本文算法重构结果在RMSE与UQI方面均优于ART与TV方法,重构图像的边缘细节亦保持良好,且抗噪性能较强,说明本文算法更适合稀疏投影情况下的重建。

Abstract: Gradient operator has some shortages in sparse representation in Computed Tomography (CT) image reconstruction algorithms, while Contourlet transform can represent image better because it contains direction information of images. Based on this, this paper introduce Contourlet transform to sparse CT image reconstruction. As a starting point, we propose an algorithm which combine Contourlet transform with Total Variation (TV) and then we use the Split Bregman method to solve the optimization problem. The experimental results show that the reconstructed results using the proposed algorithm with fewer projection can suppress noise effectively, and reduce the artifacts, which indicate that the proposed algorithm is more suitable with sparse sampling.