Asymptotics of Daubechies Filters, Scaling Functions, and Wavelets
β Scribed by Jianhong Shen; Gilbert Strang
- Publisher
- Elsevier Science
- Year
- 1998
- Tongue
- English
- Weight
- 325 KB
- Volume
- 5
- Category
- Article
- ISSN
- 1063-5203
No coin nor oath required. For personal study only.
β¦ Synopsis
We study the asymptotic form as p r Ο± of the Daubechies orthogonal minimum phase filter h p [n], scaling function f p (t), and wavelet w p (t). Kateb and Lemarie Β΄calculated the leading term in the phase of the frequency response
leads us to a problem in stationary phase, for an oscillatory integral with parameter t. The leading terms change form with t Γ t/p and we find three regions for f p (t):
(1) An Airy function up to near t 0 :
(3) A rapid decay after t 1 : (1/pp)(1/(t 0 t 1 ))sin[p(G ( 01) (p) 0 tp)] / o(p 01 ) The numbers t 0 Γ‘ 0.1817 and t 1 Γ‘ 0.3515 are known constants. The function G and its integral G ( 01) are independent of p. Regions 1 and 2 are matched over the interval p 02/3 ΣΆ t 0 t 0 ΣΆ 1.
The wavelets have a simpler asymptotic expression because the Airy wavefront is removed by the highpass filter. We also find the asymptotics of the impulse response h p [n] -a different function g(v) controls the three regions.
The difficulty throughout is to estimate the phase.
π SIMILAR VOLUMES
We give asymptotic approximations of the zeros of certain high degree polynomials. The zeros can be used to compute the filter coefficients in the dilation equations which define the compactly supported orthogonal Daubechies wavelets. Computational schemes are presented to obtain the numerical value
This paper describes a form of discrete wavelet transform, which generates complex coefficients by using a dual tree of wavelet filters to obtain their real and imaginary parts. This introduces limited redundancy (2 m : 1 for m-dimensional signals) and allows the transform to provide approximate shi
We consider implementation of operators via filter banks in the framework of the multiresolution analysis. Our method is particularly efficient for convolution operators. Although our method of applying operators to functions may be used with any wavelet basis with a sufficient number of vanishing m