Hermitian wavelet

From testwiki
Jump to navigation Jump to search

Template:Short description Template:Multiple issues

Hermitian wavelets are a family of discrete and continuous wavelets used in the constant and discrete Hermite wavelet transforms. The nth Hermitian wavelet is defined as the normalized nth derivative of a Gaussian distribution for each positive n:[1]Ψn(x)=(2n)−n2cnHen(x)e−12x2,where Hen(x) denotes the nth probabilist's Hermite polynomial. Each normalization coefficient cn is given by cn=(n12−nΓ(n+12))−12=(n12−nπ2−n(2n−1)!!)−12n∈ℕ. The function Ψ∈Lρ,μ(−∞,∞) is said to be an admissible Hermite wavelet if it satisfies the admissibility condition:[2]

CΨ=∑n=0∞‖Ψ̂(n)‖2‖n‖<∞

where Ψ̂(n) are the terms of the Hermite transform of Ψ.

In computer vision and image processing, Gaussian derivative operators of different orders are frequently used as a basis for expressing various types of visual operations; see scale space and N-jet.[3]

Examples

The first three derivatives of the Gaussian function with μ=0,σ=1:f(t)=π−1/4e(−t2/2),are:f′(t)=−π−1/4te(−t2/2),f″(t)=π−1/4(t2−1)e(−t2/2),f(3)(t)=π−1/4(3t−t3)e(−t2/2),and their L2 norms ‖f′‖=2/2,‖f″‖=3/2,‖f(3)‖=30/4.

Normalizing the derivatives yields three Hermitian wavelets:Ψ1(t)=2π−1/4te(−t2/2),Ψ2(t)=233π−1/4(1−t2)e(−t2/2),Ψ3(t)=21530π−1/4(t3−3t)e(−t2/2).

See also

References

Template:Reflist