Answers are grounded in this paper and cite their sources.
Study introduces an intrinsic lifting scheme for quaternion graph signals, highlighting its geometric basis.
This paper introduces an intrinsic lifting scheme for quaternion graph signals built around a single geometric quantity: the geodesic correlation between rotations. The same quantity is used to construct the graph, define the coarse–detail partition, and assign prediction weights, giving these stages a common geometric basis. To support the prediction step, the canonical relative-rotation domain is formed and divided into Haar-balanced radial shells, and a cubic B-spline evaluated at the shell centers gives greater weight to neighbors that are closer to the selected reference. The predicted quaternion is then obtained directly from the relative rotation angles and axes, without mapping the data to a tangent space or computing an iterative manifold mean.
No takes yet. Share an insight, caveat, or question.
Sasan Sepehrirad (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: