基于全变差模型的自适应参数选择方法

Adaptive Parameter Selection Method Based on Total Variation Model

针对全变差去噪模型中不能自适应调整正则项参数的缺点,基于Chambolle对偶算法,提出一种自适应对偶投影算法。通过构建全变差模型中的权重与正则项和逼近项之间的函数关系,根据不动点迭代理论,得出权重的更新准则。对于模型中新引入的参数选取问题,根据多元线性回归模型拟合出新参数的选取模型,通过显著性检验,验证了该拟合模型的有效性。实验结果表明,该算法能有效改善去噪效果。

As for the disadvantage of not being able to adaptively adjust the regularization parameters in the total variation model,an adaptive dual projection algorithm was proposed based on Chambolle’s dual algorithm.With the fixed point iteration theory,the relationship among weights,regularization term and approximation term was built,and the update criterion of weight was obtained.For the proposed parameter selection problem,the multiple linear regression model was used to fit the selection model of new parameter.The potential of the proposed method was verified through the significance test.The experimental results show that the algorithm is effective in improving the denoising effect.