VideoGCS - A Clustering-Based System for Video Summarization and Browsing

We present a new approach for MPEG-2 video summarization and browsing based on the Growing Cell Structures (GCS) neural algorithms. Our system VideoGCS first applies GCS to select keyframes for each shot using the color histograms of the dc-images of I frames. These keyframes are then clustered by TreeGCS to form a hierarchical view of the video for efficient browsing. GCS is able to capture very well the video content and outperforms two other approaches. The main advantage of the TreeGCS module is the ability to dynamically form a flexible hierarchy depending on the video content.