基于快速亚像素运动估计的肺4D-CT图像超分辨率重建

摘要

Super-resolution image reconstruction techniques play an important role for improving image resolution of lung 4D-CT. We presents a super-resolution approach based on fast sub-pixel motion estimation to reconstruct lung 4D-CT images. A fast sub-pixel motion estimation method was used to estimate the deformation fields between "frames", and then iterative back projection (IBP) algorithm was employed to reconstruct high-resolution images. Experimental results showed that compared with traditional interpolation method and super-resolution reconstruction algorithm based on full search motion estimation, the proposed method produced clearer images with significantly enhanced image structure details and reduced time for computation.%肺4D-CT图像超分辨率重建技术对于提高肺4D-CT图像分辨率有重要作用。本文提出一种基于快速亚像素运动估计的超分辨率处理方法，以重建清晰的肺4D-CT图像。首先采用快速亚像素运动估计方法估计不同“帧”图像间的运动场，然后采用迭代反投影方法重建高分辨率肺4D-CT图像。实验结果表明，与传统的插值方法及基于全搜索运动估计的超分辨率重建算法相比较，本方法能得到更清晰的肺部图像，图像结构明显增强。同时，与全搜索算法相比，运算时间显著缩减。