Multidimensional Fractional Wavelet Transforms and Uncertainty Principles

Let \psi,\phi\in L^2(\mathbb{R}^{N}) be two wavelets and for any two functions f,g \in L^2(\mathbb{R}^{N}). Also let \sigma>0 and r,s\in \mathbb{C}, then

1. Linearity: W^{\bm\alpha,\lambda}_{\psi}(r{f}+s{g})=r(W_{\psi}^{\bm\alpha,\lambda}{f})+s(W_{\psi}^{\bm\alpha,\lambda}{g}). This property shows that the MFrWT upholds the superposition principle, which is useful for multicomponent signal analysis.

2. Anti-linearity: W_{\left(r{\psi}+s{\phi}\right)}^{\bm\alpha,\lambda}{f}=\overline{r}W_{{\psi}}{f}+\overline{s}W_{\phi}{f}.

3. Dilation: (W_{\psi}^{\bm\alpha,\lambda}D_\sigma f)(\bm{a},\bm{b})=\displaystyle\frac{C'}{\sigma^{N}}W_{\psi}^{\bm\alpha,\frac{\lambda}{\sigma}}f(\sigma\bm{a},\sigma\bm{b}), where D_\sigma is a dilation operator, which is given by D_\sigma f(\bm{t})=f(\sigma \bm{t}) and C'=\sigma^{\frac{N}{2}}\overline{\left(\frac{c(\alpha_{\lambda})}{c\left(\alpha_{\frac{\lambda}{\sigma}}\right)}\right)},

4. Conjugacy: (W_{\psi}^{\bm\alpha,\lambda}\overline{f})(\bm{a},\bm{b})=\overline{W_{\overline{\psi}}^{-\bm\alpha,\lambda}f(\bm{a},\bm{b})}.

5. Parity: (W_{\psi}^{\bm\alpha,\lambda}Pf)(\bm{a},\bm{b})=W_{\psi}^{\bm\alpha,\lambda}f(-\bm{a},-\bm{b}), Here, P is the parity operator, which is defined by Pf(\mathbf{x})=f(-\mathbf{x}).

6. Translation: (W_\psi^{\bm\alpha,\lambda}\tau_{\mathbf{y}}f)(\bm{a},\bm{b})=e_{\bm{\alpha},-\lambda^2}(\bm{b})e_{\bm{\alpha},\lambda^2}(\bm{b}-\mathbf{y})e_{\bm{\alpha},\lambda^2}(\mathbf{y})(W_\psi^{\bm\alpha,\lambda}\check{f})(\bm{a},\bm{b}-\mathbf{y}), where fractional translation operator \tau_{\mathbf{y}} is given by \tau_{\mathbf{y}}f(\mathbf{x})=f(\mathbf{x}-\mathbf{y}) and \displaystyle \check{f}(\bm{t})=f(\bm{t})e^{i\lambda^2\sum_{k=1}^{N}a(\alpha_k)2{t_{k}}y_k}.

(FrFT of \mathbf{\Psi}_{\bm{\alpha},\lambda,\bm{a},\bm{b}}(\bm{t})) Suppose a non-zero function \psi\in L^2(\mathbb{R}^{N}) is given, then

(FrFT of W_{\psi}^{\bm\alpha,\lambda}{f}(\bm{a},\bm{b})) Suppose \psi be an admissible wavelet and for any arbitrary function f\in L^2(\mathbb{R}^{N}), then

where F(\bm{\xi})=e_{\bm{\alpha},\lambda^2}(\bm{a}\bm{\xi})(\mathfrak{F}_{\bm{\alpha},\lambda}f)(\bm{\xi})\overline{(\mathfrak{F}_{\bm{\alpha},\lambda}\psi)(\bm{a}\bm{\xi})}.

(Inner product relation for MFrWT) Let W_{\psi}^{\bm\alpha,\lambda}{f} and W_{\psi}^{\bm\alpha,\lambda}{g} represent the MFrWT of functions f and g with respect to admissible wavelet \psi respectively and then

where

Suppose \psi be an admissible wavelet and for any arbitrary function f\in L^2(\mathbb{R}^{N}). Then

(Reconstruction formula) Suppose there exist an arbitrary function f\in L^2(\mathbb{R}^{N}) and an wavelet admissible function \psi. Also let C_{\bm{\alpha},\lambda}\neq 0 as defined in equation (3.2). The reconstruction formula for f is then as follows:

(Reproducing kernel for the range of MFrWT) Let \psi be a wavelet. Also let C_{\alpha,\lambda}\neq 0 as defined in equation (3.2) and (\bm{a}_0 , \bm{b}_0) \in \mathbb{R}^{N}\times \mathbb{R}^{N}_0. Then, F\in L^2\left(\mathbb{R}^{N}\times \mathbb{R}^{N}_0, \frac{d\bm{a}d\bm{b}}{|\bm{a}|_m^2}\right) be the MFrWT of some f \in L^2(\mathbb{R}^{N}) iff

where K_{\bm{\alpha},\lambda,\mathbf{\Psi}}(\bm{a}_0,\bm{b}_0:\bm{a},\bm{b}) is the reproducing kernel which satisfies,

Let \psi be an admissible wavelet and for any function f\in L^2(\mathbb{R}^{N}). Then,

where C_{\bm\alpha,\lambda} is as defined in equation (3.2).

(Heisenberg’s inequality of uncertainty for MFrWT) Let \psi be an admissible wavelet and for any function f\in L^2(\mathbb{R}^{N}), the MFrWT of f with respect to the variable \bm\alpha is specified by (W_{\psi}^{\bm\alpha,\lambda}{f})(\bm{a},\bm{b}) . Then

where C_{\bm\alpha,\lambda} and P_{\bm\alpha,\lambda} are defined in equation (3.2) and equation (?) respectively.

(Logarithmic uncertainty inequality for MFrWT) Let \psi be an admissible wavelet and for any f\in L^2(\mathbb{R}^{N}) and for some M> 0, Let MFrWT of f with parameter \bm{\alpha} is given by (W_{\psi}^{\bm{\alpha}, \lambda}f)(\bm{a},\bm{b}). Then

where P'_{\bm\alpha,\lambda} is given by equation (?).

(Local uncertainty inequality for MFrWT) If \psi is an admissible wavelet and \theta be such that 0 < \theta < \frac{N}{2}. Then, for every measurable subsets K of \mathbb{R}^{N}, there exist constant A_\theta and for all f\in L^2(\mathbb{R}^{N}) such that

(Local uncertainty inequality for MFrWT ) Assume \psi is an admissible wavelet and \theta is such that \theta \textgreater \frac{N}{2}. Then, for every measurable subset K of \mathbb{R}^{N}, there exist a constant A_\theta and for all f\in L^2(\mathbb{R}^{N}) such that