Optimally Sparse Representations using Shearlets
Kanghui Guo, Demetrio Labate
Abstract:
It is now widely acknowledged that traditional wavelets are not
very efficient in dealing with multidimensional signals containing
distributed discontinuities. In this talk, we describe
a new multiscale directional representation system, called the
shearlet representation. This approach, which is based on
the theory of composite wavelets recently introduced by the author
and his collaborators, combines the power of multiscale methods
with a unique ability to capture the geometry
of multidimensional data. Indeed, the shearlet representation provides
optimal approximations for 2-dimensional functions with smooth
discontinuities. Numerical experiments demonstrate
that this approach has great potential in several image processing
applications.
|
