Modelling Video-Quality Shaping with Interpolation and Frame-Drop Patterns

This paper investigates combining video frames of different spatial resolutions to model and study sub-object dropping based on the concept of Quality Shaping in the Distributed Multimedia Plays (DMP) architecture for future networked collaboration. It is aimed at reducing the data rate needed to stream video data over a network while keeping the perceived video quality as high as possible. Frame-drop patterns are introduced for this purpose and two different patterns were examined using two different temporal resolutions and using different frame-drop patterns and spatial resolution on the low quality frames. The results are assessed in a qualitative manner with an action-research approach. The experimental results show that different frame-drop patterns and spatial resolution on low quality frames can be combined to have a large impact on the perceived quality of the resulting video sequence. Some combinations can give a perceived quality almost as good as the original one without any frame drop or resolution reduction, saving network bandwidth considerably.