基于非线性多重网格的TV^(p)配准  

A TV^(p) Registration Model Based on Nonlinear Multigrid

作　　者：韩晓花 杨奋林[1] HAN Xiaohua;YANG Fenlin(School of Mathematics and Statistics,Jishou University,Jishou 416000,Hunan,China)

机构地区：[1]吉首大学数学与统计学院,湖南吉首416000

出　　处：《山西师范大学学报（自然科学版）》2022年第3期39-44,共6页Journal of Shanxi Normal University(Natural Science Edition)

基　　金：国家自然科学基金资助项目(11501243);湘西自治州科技项目(2018SF5021);吉首大学校级科研项目(Jdy20060).

摘　　要：设计了一种简化的p函数以简化TV^(p)配准模型,以滞后不动点迭代和Gauss⁃Seidel松弛迭代相结合构造光滑化方法,给出了一种有效的非线性多重网格(NMG)算法.数值实验表明该算法与不动点迭代方法(FP)相比具有更好的配准速度和配准精度.Based on the multigrid strategy,a fast registration algorithm(NMG)is designed to solve the simplified TV^(p) registration model which preserves the discontinuity and smoothness of displacement field.The algorithm combines fixed point iteration and Gauss Seidel relaxation iteration as smoothing method,Numerical experiments show that the proposed algorithm has better registration speed and accuracy than fixed point iterative method(FP).

关 键 词：图像配准 全变分 自适应 非线性多重网格算法 Gauss⁃Seidel松弛迭代 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]