The structure of a reconstruction set (the set of all signals satisfying a given representation) is studied. Assuming finite data length and using the finite-dimensional linear space approach, a general form of a reconstruction set from a discrete wavelet maxima representation is presented. Necessary and sufficient conditions for uniqueness are described. These conditions can be implemented as a test for uniqueness. Although, in most cases, the discrete wavelet maxima representation is unique, the conjecture about the uniqueness of the wavelet maxima representation is not true. A family of sequences with the same maxima representation is shown. The exact reconstruction set is calculated, and it is shown to consist of similarly shaped sequences.<<ETX>>
[1]
Stéphane Mallat,et al.
A Theory for Multiresolution Signal Decomposition: The Wavelet Representation
,
1989,
IEEE Trans. Pattern Anal. Mach. Intell..
[2]
Andrew P. Witkin,et al.
Scale-Space Filtering
,
1983,
IJCAI.
[3]
Stéphane Mallat,et al.
Zero-crossings of a wavelet transform
,
1991,
IEEE Trans. Inf. Theory.
[4]
Andrew P. Witkin,et al.
Scale-space filtering: A new approach to multi-scale description
,
1984,
ICASSP.
[5]
Robert Hummel,et al.
Reconstructions from zero crossings in scale space
,
1989,
IEEE Trans. Acoust. Speech Signal Process..