An interactive method for dynamic video synopsis generation

In this paper, we propose an interactive method to generate dynamic video synopsis by integrating both content and context information. First, the methods for content and context information mining are presented. Then, we construct a network with both cues for video shots correlation. Moreover, the mechanism for dynamic video synopsis generation with three different strategies is presented. Based on the analysis video structure and content, we build a system for dynamic video synopsis generation. Different from previous work, our research focuses on two aspects: (1) dynamic synopsis which replaces fixed video summary; (2) dynamic recommendation scheme which replaces retrieval technique. Promising results demonstrate that our approach can discover the latent relationship among video shots with video saliency extraction, semantic annotation and user preference analysis. Therefore, the proposed framework can effectively lead viewers to preview novel movies in individual manner.