Generalized Inverses of Quaternion Matrices with Applications (2506.19308v1)

Abstract: This study investigates the theoretical and computational aspects of quaternion generalized inverses, focusing on outer inverses and {1,2}-inverses with prescribed range and/or null space constraints. In view of non-commutative nature of quaternions, a detailed characterization of the left and right range and null spaces of quaternion matrices is presented. Explicit representations for these inverses are derived, including full rank decomposition-based formulations. We design two efficient algorithms, one leveraging the Quaternion Toolbox for MATLAB (QTFM), and another employing a complex structure-preserving approach based on the complex representation of quaternion matrices. With suitable choices of subspace constraints , these outer inverses unify and generalize several classical inverses, including the Moore-Penrose inverse, the group inverse, and the Drazin inverse. The proposed methods are validated through numerical examples and applied to two real-world tasks: quaternion-based color image deblurring, which preserves inter-channel correlations, and robust filtering of chaotic 3D signals, demonstrating their effectiveness in high-dimensional setings.