基于低秩矩阵近似的低剂量CT图像两步去噪方法

Two-Step Denoising Method for Low-Dose CT Images Based on Low-Rank Matrix Approximation

低剂量CT扫描技术能有效减少患者所受到的辐射,但同时也导致了图像质量降低,尤其是图像中的条形噪声,给去噪工作带来了不小的挑战。针对这个问题,提出了一种基于低秩矩阵近似的低剂量CT图像去噪方法。该方法将低剂量CT图像的去噪过程分为两个阶段:第一阶段,利用图像结构信息的低秩特性,使用加权核范数最小化方法提取图像结构信息,实现对弱条形伪影和没有方向性的斑点噪声的去除;第二阶段,利用残留强条形伪影的旋转极低秩特性,即旋转后条纹噪声具有比背景内容更低的秩,再次利用低秩方法引入新的正则项,实现条纹噪声的去除。实验结果表明,该算法能够有效去除低剂量CT图像的条形噪声,并且较好地保留图像细节。

Low-dose CT scanning technology can effectively reduce the radiation received by patients,but it also reduces the image quality,especially the stripe noise in the image,which brings a lot of challenges to the denoising work.Aiming to solve this problem,this paper proposes a two-stage denoising method for low-dose CT images based on low-rank matrix approximation.The method divides the denoising process of low-dose CT images into two stages.In the first stage,using the low-rank characteristic of image structure information,the weighted kernel norm minimization method is used to extract the image structure information,so as to achieve the removal of weak stripe artifacts and speckle noise without directionality.In the second stage,the residual strong stripe artifact has the characteristic of extremely low rank after rotation,that is,the stripe noise after rotation has a lower rank than the background content,and the low-rank method is used again to introduce a new regular term to remove the stripe noise.The experimental results show that the algorithm can effectively remove the stripe noise of low-dose CT images and preserve image details well.