基于自适应NLM修正的CT重建算法

An CT Image Reconstruction Algorithm Based on Adaptive Non-Local Means Modification

针对稀疏角度CT图像重建问题,J.Huang等人提出了ART-NLM算法。该算法利用NLM对每次ART重建后的图像进行滤波去噪。虽然噪声抑制得到明显改善,但图像结构边缘模糊。论文在ART-NLM基础上改进NLM,通过自适应选取滤波参数及相似块的旋转变换,对待重建图像进行自适应NLM滤波,以达到同时去除噪声和保护边缘的目的。Shepp-Logan体模仿真实验表明,与传统ART-NLM算法相比,新算法提高了重建图像质量,有效平衡了图像去噪和边缘保护。

For the problem of sparse angular CT image reconstruction, J. Huang proposed ART - NLM algorithm. In one iteration, traditional NLM is used to reduce the noise of the image reconstructed by ART algo- rithm. The noise in the reconstructed image can be suppressed well, but the edges are often overly smoothed. In this paper, this algorithm was improved with adaptive NLM. In adaptive NLM, adaptive filtering parameter selection and rotation transform of similarity patch are proposed to reduce noise and preserve edges simultaneously. The simulation results of Shepp - Logan show that this new method improves the reconstructed image quality effectively and achieves a good balance between noise suppression and edges preservation, compared to conventional ART - NLM.