Video coding infrastructure using adaptive prediction complexity reduction

Abstract

Within a device, one or more subsets associated with one or more frames or pictures of the video signal may be adaptively selected and used for motion vector calculation (e.g., such as in accordance with inter-prediction). For example, a picture or frame of the video signal may be partitioned into a number of respective regions. Any one or more, but typically fewer than all, of the respective regions of that picture or frame may be appropriately selected, and stored, based on any one or more considerations for use in motion vector calculation (e.g., inter-prediction). A sub-sampled or down-sampled picture or frame [or alternatively, a sub-sampled or down-sampled version of one or more respective regions of a picture or frame] (e.g., the sub-sampling or down-sampling ratio which may be adaptively determined based on any one or more considerations) may be stored for use in motion vector calculation (e.g., inter-prediction).