Traditional methods on video summarization are designed to generate summariesfor single-view video records; and thus they cannot fully exploit theredundancy in multi-view video records. In this paper, we present a multi-viewmetric learning framework for multi-view video summarization that combines theadvantages of maximum margin clustering with the disagreement minimizationcriterion. The learning framework thus has the ability to find a metric thatbest separates the data, and meanwhile to force the learned metric to maintainoriginal intrinsic information between data points, for example geometricinformation. Facilitated by such a framework, a systematic solution to themulti-view video summarization problem is developed. To the best of ourknowledge, it is the first time to address multi-view video summarization fromthe viewpoint of metric learning. The effectiveness of the proposed method isdemonstrated by experiments.
展开▼