Correlation-aware packet scheduling for multi-camera streaming
In multi-view applications, multiple cameras acquire the same scene from different perspectives, which results in correlated video streams. It becomes important to exploit this correlation at the acquisition side (i.e., in the source coding) or at the receiver side (i.e., during error-concealment). In this work, we propose a correlation-aware scheduling algorithm for multi-camera sets, in which information from all views need to be sent over a bottleneck channel to clients that decode the 3D scene captured by the cameras. Based on a novel rate-distortion model, that takes into account the correlation between sources, we propose a solution that minimizes the distortion in the scene reconstruction and adapts to temporal variations in the scene content. Simulation results show the gain of the scheduling algorithm when the correlation model is known in the optimization, compared to scheduling policies with no information about the correlation or with a priori camera selection algorithms.