Chroma interpolation using windowed kriging for color-image compression-by-network with guaranteed delay

Artistic elements in immersive collaborations, like collaborative percussions and dancing, depend on visual cues and thus require a same constant delay for both audio and video data to ensure harmonious synchronization. We envision the compression-by-network approach where network nodes can reduce video data intelligently due to immediate traffic conditions to guarantee maximal constant delay for video data with graceful quality degradation. To process color information with that approach, in this paper we propose using windowed kriging for chroma interpolation and investigate the effects. We define color compressibility as the maximal compression ratio of color information without noticeable artifact. Our results show that, based on compressibility, network nodes can discard 99% of chroma pixels and the receiver interpolates the chroma images using only the received 1% to yield the output color image without salient artifact. We discover that red and similar colors have lower compressibility than blue or green.