METHOD AND ENCODER SYSTEM FOR PACKET-CONSTRAINED SPECTRAL PREPARATION OF THREE-DIMENSIONAL VOLUMETRIC DATA FOR REAL-TIME STREAMING

The invention relates to a computer-implemented method, encoder system, and computer-readable medium for preparing three-dimensional volumetric data for real-time streaming over a packet network under packet-payload constraints. The method represents the volumetric data as a weighted discrete spatial graph including boundary nodes and bulk nodes, and derives a boundary state representation including a two-dimensional boundary matrix. For each packet to be generated, a maximum payload length is determined from a transmission-unit constraint and header overhead. Selected boundary elements are quantised and projected into a graph spectral domain defined by a weighted graph Laplacian, and projected components above a dynamically determined spectral cut-off are discarded before serialisation. The cut-off threshold, or a corresponding spectral truncation index, is updated as a function of the available payload budget and at least one measured link metric, such as bitrate, packet loss, latency, or jitter, so that retained data fits within the packet payload budget without iterative trial-and-error truncation. In optional embodiments, a safety margin dependent on jitter is used, and the truncation rule may be implemented by closed-form computation or lookup tables.