Abstract:Based on the discrete fractional random transform (DFRNT), we present the discrete fractional random cosine and sine transforms (DFRNCT and DFRNST). We demonstrate that the DFRNCT and DFRNST can be regarded as special kinds of DFRNT and thus their mathematical properties are inherited from the DFRNT. Numeral results of DFRNCT and DFRNST for one and two dimensional functions have been given.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to propose the discrete fractional - order random cosine transform (DFRNCT) and the discrete fractional - order random sine transform (DFRNST) based on the discrete fractional - order random transform (DFRNT), and to study their mathematical properties and their relationships with DFRNT. Specifically, the author hopes that these newly proposed transforms can inherit the excellent mathematical properties of DFRNT and serve as general - purpose mathematical and numerical tools in the field of digital signal processing.
### Background and Motivation of the Paper
1. **Proposal and Application of DFRNT**:
- The author previously proposed the discrete fractional - order random transform (DFRNT) and demonstrated its application in image encryption and decryption.
- DFRNT is derived from the fractional - order Fourier transform (FrFT) and the discrete fractional - order Fourier transform (DFrFT), so it has excellent mathematical properties, and at the same time, its transformation results are inherently random.
2. **Extension to DFRNCT and DFRNST**:
- Due to the close relationship between DFRNT and DFrFT (FrFT), the author hopes that DFRNT can serve as a general - purpose tool in the field of digital signal processing.
- Therefore, the author further extends DFRNT to the discrete fractional - order random cosine transform (DFRNCT) and the discrete fractional - order random sine transform (DFRNST). These transforms have a wide range of applications in signal and image processing, such as signal coding, watermark embedding, and blurred image restoration.
### Main Contributions
1. **Definition of DFRNCT and DFRNST**:
- The author directly defines DFRNCT and DFRNST based on DFRNT and studies their mathematical properties.
- It is found that DFRNCT and DFRNST can be regarded as special cases of DFRNT, so they inherit the mathematical properties of DFRNT.
2. **Study of Mathematical Properties**:
- The properties such as linearity, unitarity, exponential additivity, and periodicity of DFRNCT and DFRNST are studied.
- The inverse transformation formulas of DFRNCT and DFRNST are proposed.
3. **Reconstruction of DFRNT (ReDFRNT)**:
- By combining the eigenvectors and eigenvalues of DFRNCT and DFRNST, the author reconstructs a new DFRNT (ReDFRNT).
- ReDFRNT has good symmetry for dual signals and is suitable for the calculation of even and odd signals.
### Numerical Results
- The author verifies the symmetry and consistency of DFRNCT, DFRNST, and ReDFRNT through numerical simulation.
- The results show that ReDFRNT maintains the symmetry of the input image when processing two - dimensional functions.
### Application Prospects
- This research provides a theoretical basis and the possibility of practical application for DFRNT, DFRNCT, DFRNST, and ReDFRNT in fields such as information security (such as image encryption and decryption), image processing, pattern recognition, and artificial intelligence.
In summary, this paper aims to further expand the application range of DFRNT by introducing DFRNCT and DFRNST and to provide new mathematical tools for the field of digital signal processing.