Randomized low rank approximation for nonnegative pure quaternion matrices (Q6202721)

This interesting paper deals with randomized low rank approximations for nonnegative pure quaternion matrices. It is well known that pure quaternion matrices are applied widely to represent real-world data in various applications such as colour images and fluorescent images, whose pixel values of different bands can be encoded on the imaginary parts of the matrix. Moreover, the imaginary part of the given quaternion matrix is usually nonnegative due to the natural nonnegativity of real-world data. In this regard, low rank quaternion matrix approximation is an important problem in several quaternion application fields. See the many references in the paper under review for some of these interesting fields. The authors, propose an alternating projection-based algorithm for low rank nonnegative pure quaternion matrix approximation, which can exactly calculate an optimal fixed rank approximation while at the same time, preserve the pure and nonnegative properties from the given data. The proposed algorithm alternatively projects the given quaternion matrix onto a fixed rank quaternion matrix set and nonnegative pure quaternion matrix set in an iterative fashion. The authors study theoretical convergence guarantees of their algorithm. To extend their proposed algorithm to large-scaled data, the authors propose a randomized algorithm with significant lower computational complexity and comparable accuracy. Numerical experiments performed by the authors on colour images show that their algorithms outperform the other state-of-the-art algorithms. The paper is well written with a good set of references.