Abstract: On the base of the classic total variation de-noising model and its improved models, this paper proposes an improved double-entry adaptive Total Variation(TV) model based on local information of pre-made images. An adaptive spatial fidelity term is intended to ease the smoothing effect by the second-order nonlinear filtering over the details. An adaptive regularization term is to further reduce the staircase effect, also to achieve a more stable and convergence value. Experimental results show that the improved method still can achieve better results in high noise. Compared with the original method, it has better noise robustness.

Key words: image de-noising, staircase effect, adaptive fidelity term, adaptive regularization term, residual image, average gradient

摘要： 研究经典总体变差去噪模型及其改进的自适应去噪模型，提出一种基于预处理图像局部信息的双项自适应模型。利用空间自适应保真项缓解二阶非线性滤波对细节的过度平滑，通过自适应正则化项减少阶梯效应，使数值更稳定和收敛。实验结果表明，与原方法相比，改进方法具有更好的鲁棒性，在噪声较高的情况下仍能取得较好的去噪效果。

关键词: 图像去噪, 阶梯效应, 自适应保真项, 自适应正则化项, 残差图像, 平均梯度

CLC Number: 

• TN911.73