Unsupervised Clustering by k-medoids for Video Summarization

In this paper, we propose a video summarization algorithm by multiple extractions of key frames in each shot. This algorithm is based on the k partition algorithms. We choose the ones based on k-medoid clustering methods so as to find the best representative object for each partitions. In order to find the number of partition (i.e. the number of representative frames of each shot), we introduce a quantity based on the distance between frames and on the size of the video shot. This algorithm, which is applicable to all types of descriptors, consists of extracting key frames by similarity clustering according to the given index (histogram features, motion features, texture features, or a combination of these features). In our proposal, the distance between frames is calculated using a fast full search block matching algorithm based on the frequency domain. The proposed approach is computationally tractable and robust with respect to sudden changes in mean intensity within a shot. Additionally, this approach produces different key frames even in the presence of a large motion. The experiment results show that our algorithm extracts multiple representative frames in each video shot without visual redundancy, and thus it is an effective tool for video indexing and retrieval.