Parameterization of windowed kriging for compression-by-network of natural images

Artistic elements in immersive collaborations that depend on visual cues, such as collaborative percussions and dancing, require a maximal end-to-end delay for both audio and video data to ensure harmonious synchronization. Such delays have been shown to be extremely low in milliseconds. To guarantee such delay for video data with graceful quality degradation, we envision that network nodes can reduce video data intelligently in changing traffic conditions. Thus we call this approach compression-by-network in contrast to the source coding that comes with no such guarantee. Windowed kriging interpolation has been introduced as a technique that matches the novel approach. This paper presents an empirical study on the parameters of the technique for interpolating grayscale natural images in terms of image quality and processing time. Recommended values for the parameters are given together with interesting insights as the basis to extend the application of the technique for interpolating chroma images.