VKCG

Video Key frame Extraction through Canonical Correlation Analysis and Graph Modularity

Abstract: Key frame based video summarization has emerged as an important area of multimedia research in recent times. In this paper, we propose a novel automated approach for key frame extraction using canonical correlation analysis (CCA) and graph modularity clustering in compressed domain. We prune certain edges from the Video similarity graph (VSG) using an iterative strategy until there is no improvement in graph modularity. Resulting connected components in the final VSG correspond to the separate clusters. The proposed algorithm also uses multi-feature fusion using CCA to achieve higher semantic dependency between different video frames. Experimental results on some standard videos of different genre clearly indicate the superiority of the proposed method in terms of the F₁ measure.

Evaluation Dataset Information

Results: