Video Summarization by Efficient Clustering of Compressed Chromaticity Signatures

Motivated by problems in video summarization caused by change in lighting, we develop a new low-dimensional feature video frame feature that is more insensitive to such changes than previous methods. Because we effectively reduce any video to the same lighting conditions, we can produce a universal basis on which to project video frame features. We set out a new multi-stage hierarchical clustering method that merges clusters based on variance ratio, the ratio of intraand total variance, and merges only adjacent clusters. A second stage merges clusters incorrectly split in the first round by the greedy hierarchical algorithm, and finally merges non-adacent clusters to fuse near-repeat shots. The new summarization method produces a very succinct set of keyframes for videos, and results are very encouraging.