Abstract
Adaptive video streaming over HTTP has gained a significant value due to its ability to provide high Quality of Service (QoS). Also, video summarization is useful when there are tight constraints (e.g. low available bandwidth, low energy level, low available time for watching, etc.). Therefore, this paper presents a Context-aware joint Video Summarization and Streaming (CVSS) approach for mobile devices. This approach utilizes the flexibility of Dynamic Adaptive Streaming over HTTP (DASH) to provide an adaptive video summary. The video summary (dynamic) is generated based on the Visual Attention Model (VAM) and a new Fast Directional Motion Intensity Estimation (FDMIE) algorithm. The adaptation process is based on various contexts, which include network context, device context and user context. A client/server prototype was implemented to evaluate the CVSS approach. The evaluation results demonstrate that:
1. The video summary generated by CVSS has an effectiveness rate up to 87% with respect to the manually generated summary and the state of the art approaches. Moreover, the efficiency of the proposed summarization approach makes it suitable for online and mobile applications.
2. The average initial delay is 0.62 second, the average jitter is 0.43 second and the average packet loss ratio is less than 1.4 %. These results show that, CVSS approach greatly achieves the standard QoS requirements for on-demand video streaming.