基于四元数的边缘自适应正则化模型  

Quaternion Based Edge Adaptive Regularization Model

作　　者：贺志飞 曹军[1] HE Zhifei;CAO Jun(School of Mathematical Sciences,Inner Mongolia University,Hohhot 010021,China)

机构地区：[1]内蒙古大学数学科学学院,呼和浩特010021

出　　处：《内蒙古大学学报（自然科学版）》2022年第3期253-260,共8页Journal of Inner Mongolia University：Natural Science Edition

基　　金：内蒙古自然科学基金(2020MS01002)。

摘　　要：通过在HSV空间的全变分模型中加入了边缘自适应约束,得到基于四元数的边缘自适应全变差模型。该模型利用HSV空间各通道的特点得到图像的边缘信息,然后自动调节选择适当的参数作为边缘约束,解决之前模型恢复图像的边缘模糊或细节缺失问题。证明了算法的全局收敛性,并使用ADMM求解。数值实验表明了提出方法的有效性。An edge adaptive total variation model based on quaternion is obtained by adding edge adaptive constraints to the total variation model in HSV space.The model uses the characteristics of each channel in HSV space to obtain the edge information of the image,and then automatically adjusts and selects appropriate parameters as edge constraints to solve the problem of edge blur or missing details of the image restored by the previous model.The global convergence of the algorithm is proved and solved by ADMM.Numerical experiments show the effectiveness of the proposed method.

关 键 词：四元数 ADMM 全变差 图像恢复 

分 类 号：O29[理学—应用数学]