Contourlet-Based Tree Approximation and Encoding

A new tree approximation based on contourlet transform is proposed in this paper. Contourlet transform produce a simple and effective representation frame for geometrical image by employing Laplace-pyramid decomposition and followed by direction filter bank (DFB). However, the redundancy of contourlet constrains its application in image representation. Fortunately, as one of many nonlinear approximation methods, tree structure poses an economical encoding method for images. Because of the refinement of tree structure for nonlinear approximation, a tree structure encoding method for image based on contourlet transform is also introduced. Experiments show that contourlet-based tree approximation outperforms wavelet-based tree approximation.