Abstract: The Total Variation（TV） regularization model is a classical denoising model, which can be solved by the split Bregman algorithm simply and effectively. In this paper, an improved anisotropic total variation denoising model based on the TV-L1 model and the OSV model is proposed. It is also solved by the split Bregman algorithm. Numerical results show that the improved model can protect the edges, highlight the geometry features and texture, make the denoising image become more clear and result in a better image denoising effect.

Key words: image denoising, TV model, anisotropic total variation, split Bregman iteration

摘要： 全变差（TV）正则化模型是最经典的去噪模型，利用分裂的Bregman迭代算法可以简单有效地求解该模型。结合TV-L1模型和OSV模型，提出了一种改进的各向异性全变差去噪模型，并且利用分裂的Bregman迭代算法进行求解。通过数值实验可以看出改进的模型保护了恢复图像的边缘，突出了几何特征和纹理，使其更加清晰，去噪效果比原模型有所提高。

关键词: 图像去噪, TV模型, 各向异性全变差, 分裂的Bregman迭代