A generalized wavelet extrema representation
Conference
·
OSTI ID:225035
The wavelet extrema representation originated by Stephane Mallat is a unique framework for low-level and intermediate-level (feature) processing. In this paper, we present a new form of wavelet extrema representation generalizing Mallat`s original work. The generalized wavelet extrema representation is a feature-based multiscale representation. For a particular choice of wavelet, our scheme can be interpreted as representing a signal or image by its edges, and peaks and valleys at multiple scales. Such a representation is shown to be stable -- the original signal or image can be reconstructed with very good quality. It is further shown that a signal or image can be modeled as piecewise monotonic, with all turning points between monotonic segments given by the wavelet extrema. A new projection operator is introduced to enforce piecewise inonotonicity of a signal in its reconstruction. This leads to an enhancement to previously developed algorithms in preventing artifacts in reconstructed signal.
- Research Organization:
- Lawrence Livermore National Lab., CA (United States)
- Sponsoring Organization:
- USDOE, Washington, DC (United States)
- DOE Contract Number:
- W-7405-ENG-48
- OSTI ID:
- 225035
- Report Number(s):
- UCRL-JC--122393; CONF-960672--2; ON: DE96004156
- Country of Publication:
- United States
- Language:
- English
Similar Records
A wavelet phase filter for emission tomography
Scale-dependent monotonicity and scale-dependent extrema
Wavelet Representation of Contour Sets
Technical Report
·
Sat Jul 01 00:00:00 EDT 1995
·
OSTI ID:95206
Scale-dependent monotonicity and scale-dependent extrema
Conference
·
Fri Dec 30 23:00:00 EST 1994
·
OSTI ID:35863
Wavelet Representation of Contour Sets
Conference
·
Thu Jul 19 00:00:00 EDT 2001
·
OSTI ID:15006196