Technical Program

Paper Detail

Paper:	IMDSP-L7.3
Session:	Feature Extraction and Analysis
Time:	Thursday, May 18, 10:40 - 11:00
Presentation:	Lecture
Topic:	Image and Multidimensional Signal Processing: Feature Extraction and Analysis
Title:	Shift-invariant Multiscale Multidirectional Image Decomposition
Authors:	Truong T. Nguyen, Soontorn Oraintara, University of Texas, Arlington, United States
Abstract:	This paper presents a shift-invariant complex directional pyramid transform, which is implemented by a dual-tree of pyramidal directional filter banks (PDFB). The double filter bank framework consists of a shift-invariant Laplacian pyramid and a dual-tree DFB. The two DFBs (primal DFB and dual DFB) constructed by a binary tree structure are identical except for the filter banks at the second level of the dual DFB, where a special condition on the phase of the filters are required. By viewing the two subband coefficients as the real and imaginary parts of a complex-valued subband image, it is shown that there is no aliasing in the decimated complex-valued signal, which also means that the system is shift-invariant in the energy sense. The proposed shift-invariant, multiscale, multidirectional image decomposition has two unique characteristics that other shift-invariant decompositions do not possess. First, the directional resolution of the image transform is not fixed, and can be arbitrarily increased. Secondly, the two-dimensional filter bank can be implemented in a separable fashion, which makes the entire structure very computational efficient.