Temporal-domain multiple video descriptors using motion-compensated temporal filtering

Abstract

A temporal approach to multiple description video coding is proposed. The directionality of the motion compensation and temporal filtering is manipulated at different temporal levels to generate different low-frequency temporal bands. Subsets of the latter bands at various temporal levels are treated as single temporal descriptors. The proposed solution, the resultant excessive bitrate and the associated quality degradations are developed. It is shown that the use of all descriptors results in a moderate increase in bitrate yet video at the same quality of single layer coding is reconstructed. It is also shown that the use of a subset of the descriptors results in bitrate savings, albeit at a reasonable degradation in image quality.