GPU梯度自适应SART图像重建算法

Gradient Adaptive SART Image Reconstruction Algorithm Based on GPU

针对SART迭代重建算法所需投影数据量大且迭代时间长的问题,将TV算法引入SART中,动态调节梯度步长来加速算法的收敛性能,实现用少量投影数据重建出高质量的图像。为加快算法的执行速度,将SART-TV算法在GPU上并行加速。用SART算法和SART-TV算法对Shepp-Logan模型重建,仿真实验表明:在采样数据相同的情况下,SART-TV算法较SART算法可以重建出更好的图像,且基于GPU的SART-TV算法的迭代时间明显低于基于CPU的SART-TV算法。

Considering the large amount of required projection data and time -consuming of SART iterative reconstruction, adaptive step TV algorithm is introduced into the SART algorithm, which can achieve high quality image reconstruction with a small amount of projection data and increase the convergence speed of the algorithm. And SART - TV algorithm is accelerated by using GPU. Shepp - Logan model is reconstructed by SART -TV algorithm and SART algorithm respectively, and simulation experiments show that SART- TV algorithm earl reconstruct high quality image, which is better than the SART algorithm via equal sampling point, and iterative time of SART -TV algorithm based on GPU is significantly lower than that based on CPU.