VISCOM: A robust video summarization approach using color co-occurrence matrices

Marcos Vinicius Mussel Cirne, Hélio Pedrini

Abstract: Video summarization techniques have allowed the content analysis of large volumes of digital video sequences of different categories, such as movies, documentaries, lectures, sports, surveillance, and news. This paper proposes and evaluates a novel video summarization approach called VISCOM, which is based on color co-occurrence matrices to describe the video frames and generate a synopsis with the most representative frames. Experiments conducted on two different data sets of various genres demonstrate the effectiveness of the proposed method in terms of quality. The resulting video summaries are compared against several others using a specific quantitative evaluation metric, producing competitive outcomes among the evaluated methods.