Sparse Geometric Image Representations with Bandelets英文文献.pdfVIP

Sparse Geometric Image Representations with Bandelets 1 1,2,* Erwan Le Pennec , Stephane´ Mallat 1 Centre de Mathematiques´ Appliquees,´ Ecole Polytechnique, 91128 Palaiseau Cedex, France, Tel : 33 1 69 33 46 27 Fax 33 1 69 33 30 11 2 Courant Institute of Mathematical Science, New York University, New York, NY 10012, USA, email : stephane.mallat@polytechnique.fr EDICS: 1-STIL Still Image Coding; 2-NFLT Nonlinear Filtering and Enhancement; 2-WAVP Wavelets and Multiresolution Processing Abstract This paper introduces a new class of bases, called bandelet bases, which decompose the image along multiscale vectors that are elongated in the direction of a geometric flow. This geometric flow indicates directions in which the image grey levels have regular variations. The image decomposition in a bandelet basis is implemented with a fast subband filtering algorithm. Bandelet bases lead to optimal approximation rates for geometrically regular images. For image compression and noise removal applications, the geometric flow is optimized with fast algorithms, so that the resulting bandelet basis produces a minimum distortion. Comparisons are made with wavelet image compression and noise removal algorithms. I . IN TRODU CTION Image representations in separable orthonormal bases such as wavelets, local cosine or Fourier can not take advantage of the geometrical regularity of image structures. Sharp image transi