2005 · 40 citations · 12 references
ArtsVision ScienceComputer VisionDiagnostic ImagingDigital ImagingDigital Image ProcessingDigital Signal ProcessingImage AnalysisImage ProcessingImage ReconstructionSignal ProcessingFilter Bank DesignFilter BanksMultiresolution RepresentationNonlinear Function ApproximationVertical DecompositionWavelet AnalysisWavelet TransformWavelet Transform Analysis
We propose a new family of perfect reconstruction, non-redundant, and multiresolution geometrical image transforms using the wavelet transform in conjunction with modified versions of directional filter banks (DFB). In the proposed versions of DFB, we use either horizontal or vertical directional decomposition. Taking advantage of the wavelet transform that has efficient nonlinear approximation property, we add the important feature of directionality by applying the modified and regular DFB to the subbands of a few finest wavelet levels. This way we can eliminate a major portion of the artifacts usually introduced when DFB are used. The proposed hybrid wavelets and DFB (HWD) transform family provides visual and PSNR improvements over the wavelet and contourlet transforms.
12
Shiftable multiscale transforms
Eero P. Simoncelli, William T. Freeman, Edward H. Adelson et al. · IEEE Transactions on Information Theory · 1992 · 1.3K citations
Directional multiresolution image representations