Adaptive Image Decomposition by Improved Bilateral Filter
Liu Qiegen, Luo Jian-hua, Zhu Yuemin · International Journal of Computer Applications · 2011
This paper presents an adaptive image decomposition method based on the newly proposed texture descriptor, which combines contrast variation and structure correlation.By incorporating it into the classical edge-preserving bilateral filter, we construct an adaptive nonlinear filter capable of preserving edges and removing texture parts effectively.We derive the variational formulation of the proposed filter and explore its relationship with robust statistics.Comparative experiments against other state-of-the-art decomposition approaches show that better separation results of geometric structure from texture are obtained.