Image decomposition using adaptive second-order total generalized variation

Jianlou Xu, Xiang-Chu Feng, Yan Hao, Yu Han

Abstract: This paper proposes a new model for image decomposition which separates an image into a cartoon, consisting only of geometric objects, and an oscillatory component, consisting of textures or noise. The proposed model is given in a variational formulation with adaptive second-order total generalized variation (TGV). The adaptive behavior preserves the key features such as object boundaries and textures while avoiding staircasing effect. To speed up the computation, the split Bregman method is used to solve the proposed model. Experimental results and comparisons demonstrate the proposed model is more effective for image decomposition than the methods of the state-of-the-art image decomposition models.