The Properties and Fast Algorithm of Quaternion Linear Canonical Transform

Ye Zhang, Guanlei XuJournal of Signal and Information Processing 2018 · DOI

Abstract

The quaternion linear canonical transform (QLCT) is defined in this paper, with proofs given for its reversibility property, its linear property, its odd-even invariant property and additivity property. Meanwhile, the quaternion convolution (QCV), quaternion correlation (QCR) and product theorem of LCT are deduced. Their physical interpretation is given as classical convolution, correlation and product theorem. Moreover, the fast algorithm of QLCT (FQLCT) is obtained, whose calculation complexity for different signals is similar to FFT. In addition, the paper presents the relationship between the convolution and correlation in LCT domains, and the convolution and correlation can be calculated via product theorem in Fourier transform domain using FFT.

Introduction

In the field of signal processing, the Linear Canonical Transform (LCT) serves as a powerful generalization of several fundamental transforms, including the Fourier Transform (FT), the Fractional Fourier Transform (FRFT), and the Fresnel transform. While LCT has been extensively studied for scalar and complex-valued signals, its application to hyper-complex signals—specifically quaternions—offers significant potential for processing multi-dimensional data. Quaternions, consisting of one real and three imaginary components ($i, j, k$), are particularly adept at representing vector-valued information like color images (RGB) or electromagnetic fields without losing the inherent correlations between channels. This paper establishes the theoretical framework for the Quaternion Linear Canonical Transform (QLCT) and introduces an efficient computational method for its implementation.

Properties and Mathematical Framework

The authors rigorously define the QLCT and establish its core mathematical properties. These include:

  • Reversibility: Ensuring that signals can be transformed and accurately reconstructed.
  • Linearity and Additivity: Maintaining the fundamental behaviors expected of integral transforms.
  • Odd-Even Invariance: Describing how the transform handles signal symmetry.

A significant portion of the work is dedicated to deducing the quaternion convolution (QCV) and quaternion correlation (QCR) within the LCT domain. By establishing product theorems, the authors provide a bridge between abstract hyper-complex operations and classical physical interpretations, laying the groundwork for advanced filter design in optics and multi-channel signal analysis.

The Fast QLCT Algorithm (FQLCT)

The primary challenge in quaternion signal processing is the non-commutative nature of quaternion multiplication (where $ij \neq ji$). To overcome this, the authors developed the Fast QLCT (FQLCT) algorithm. The method involves:

  1. Decomposition: Breaking down the quaternion signal into specific complex-valued components.
  2. FFT Mapping: Mapping the QLCT operations onto 2D Fast Fourier Transforms (FFTs).
  3. Efficient Computation: Utilizing the product theorem to perform convolution and correlation as simple multiplications in the transform domain.

This approach reduces the computational burden from a direct calculation to a complexity of $O(MN \log_2 MN)$, making it practically viable for real-time applications.

Results and Complexity Analysis

Numerical simulations on discrete 2D signals demonstrate the efficiency of the FQLCT. For a signal of size $60 \times 60$, the fast algorithm is approximately ten times faster than direct computation. The complexity for general quaternion signals is shown to be $O(2MN \log_2 MN)$, which is only a small constant factor higher than standard scalar FFTs. Furthermore, the paper highlights that signal symmetries (axial or central) can be exploited to further reduce complexity by factors ranging from 2 to 8.

Limitations and Future Directions

Despite its efficiency, the QLCT currently lacks a counterpart to Parseval’s principle, meaning energy conservation is not inherently maintained in the transform domain. Additionally, the relationship between the QLCT and the Wigner-Ville time-frequency distribution remains an open question. Future research will likely focus on addressing these theoretical gaps and applying the FQLCT to specific engineering challenges such as color image denoising, watermarking, and multi-spectral data compression.

Summary updated

Abstract

The quaternion linear canonical transform (QLCT) is defined in this paper, with proofs given for its reversibility property, its linear property, its odd-even invariant property and additivity property. Meanwhile, the quaternion convolution (QCV), quaternion correlation (QCR) and product theorem of LCT are deduced. Their physical interpretation is given as classical convolution, correlation and product theorem. Moreover, the fast algorithm of QLCT (FQLCT) is obtained, whose calculation complexity for different signals is similar to FFT. In addition, the paper presents the relationship between the convolution and correlation in LCT domains, and the convolution and correlation can be calculated via product theorem in Fourier transform domain using FFT.
BibTeX
@article{zhang2018properties,
  author = {Ye Zhang and Guanlei Xu},
  title = {{The Properties and Fast Algorithm of Quaternion Linear Canonical Transform}},
  journal = {Journal of Signal and Information Processing},
  year = {2018},
  doi = {10.4236/jsip.2018.93012},
}
Figure