Sign in Subscribe
Algorithms

Quaternion Algorithm for Face Recognition Demonstrates Enhanced Execution Speed and Accurate Recognition

Strategic Insights

— 1 min read

Researchers from the University of Split have made a significant breakthrough in the field of mathematics, developing a novel quaternion algorithm for face recognition that showcases enhanced execution speed and accurate recognition. The algorithm, implemented in the Julia programming language, enables the processing of larger images and achieves comparable recognition accuracy to existing methods. This innovative approach has far-reaching implications for the field of biometrics and bioengineering.

Key Takeaways:

  • The quaternion algorithm for face recognition models images using quaternion matrices and projects them onto a carefully chosen subspace to reduce dimensionality while preserving relevant information.
  • The algorithm's effectiveness is demonstrated through numerical tests on a widely used database for face recognition, achieving comparable recognition accuracy to existing methods.
  • The implementation of the Julia programming language allows for low execution times and the capability to handle larger image dimensions.
  • The novel Jacobi method developed for solving the quaternion Hermitian eigenproblem is an essential part of the algorithm.
  • The research supports the use of quaternions as an efficient way to represent RGB images in image processing.

Statistics:

  • The algorithm reduces the dimensionality of quaternion matrices while preserving relevant information.
  • The results demonstrate a comparable recognition accuracy to existing methods.
  • The implementation in Julia allows for low execution times and the capability to handle larger image dimensions.
  • The algorithm's execution speed is enhanced due to the use of the novel Jacobi method.
  • The quaternion algorithm is suitable for processing RGB images.

Sources:

  • Mathematics, "Fast Quaternion Algorithm for Face Recognition," 2025, 13(12):1958. (http://www.mdpi.com/journal/mathematics)
  • University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, Rudjera Boskovica 32, 21000 Split, Croatia.
  • MDPI AG, publisher of Mathematics.
  • doi.org/10.3390/math13121958

Read more