Two-dimensional quaternion linear canonical transform: properties, convolution, correlation, and uncertainty principle

From MaRDI portal
Publication:2299986





A quaternionic version of the linear canonical transform (LCT) is defined. The LCT is an extension of the Fourier transform to a family of unitary integral kernel operators associated with an \(SL(2,\mathbb{R})\)-extension of the matrix \(\left(\begin{array}{cc}0 & 1\\ -1 & 0 \end{array}\right)\) to matrices \(A_s=\left(\begin{array}{cc}a_s & b_s\\ c_s & d_s \end{array}\right)\) with determinant one. When \(b_1\neq 0\) one defines the kernel \(K_{A_s}(z_s,\nu_s)=\frac{1}{\sqrt{2\pi \mu b_s}} e^{(1/2)\mu ((a_s/b_s)z_s^2-(2/b_s)z_sv_s+(d_s/b_s)v_s^2)}\) and \(K_{A_s}^\prime=\sqrt{d_1} e^{\mu (c_1d_1/2)\nu_1^2}\) (if \(b_1=0\)). One then defines the quaternionic linear canonical transform (QLCT) \[L^{\mathbb{H}}_{A_1,A_2}\{f\}(v)=\int_{\mathbb{R}^2} f(z) K_{A_1}(z_1,v_1) K_{A_2}(z_2,v_2) dz \] where both \(K_{A_i}\) are replaced by \(K_{A_i}^\prime\) if \(b_1b_2=0\). When \(A_1=A_2=\left(\begin{array}{cc}0 & 1\\ -1 & 0 \end{array}\right)\), the QLCT reduces to the quaternionic Fourier transform defined by \[\mathcal{F}_q\{f\}(v)=\int_{\mathbb{R}^2} f(z)e^{-\mu v\cdot z} \] where \(\mu\) is any pure quaternion with \(\mu^2=-1\). The inverse Fourier transform replaces the kernel \(e^{-\mu v\cdot z}\) with \(e^{\mu v\cdot z}/(4\pi^2)\). With convolution defined in the usual way as \((f\star g)(z)=\int f(y) g(z-y)\, dy\), one has \(\mathcal{F}_q(f\star g) =\mathcal{F}_q\{g\} \mathcal{F}_q\{f\}\) when \(f\) is a pure quaternionic component and, generally, it is the sum of the products over the pure quaternionic components of \(f\). An inversion property for the QLCT is established along with several covariance properties with respect to translation, modulation and dilation that are recorded in Table 1. A formula is established for a QLCT-convolution operator that partly incorporates the QLCT kernel in the integral defining convolution of \(f\) and \(g\). Finally, the following uncertainty bound is established: \[\int_{\mathbb{R}^2} z_i^2|f(z)|^2\, dz \int_{\mathbb{R}^2} v_i^2 |L^{\mathbb{H}}_{A_1,A_2} \{f\}(v)|^2 dv\geq \frac{b_i^2}{4}\|f\|^2 \] for \(i=1,2\), where equality holds precisely when \[f(z)=K_0 e^{\mu v_0\cdot z}e^{-(z_1-k_1)^2/(\alpha_1)-(z_1-k_2)^2/(\alpha_2)}\] where \(\alpha_i>0\) and \(K_0\in\mathbb{H}\) is a fixed quaternion.



Cites work









This page was built for publication: Two-dimensional quaternion linear canonical transform: properties, convolution, correlation, and uncertainty principle

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2299986)