Basic premises for a novel method development are observed, which is used for video content analyses based on Voronoi diagrams. Some peculiarities are derived from the performed experiments. These peculiarities are connected with an amount and location of salient points that are used for Voronoi diagram construction. Higher order Voronoi diagram implementation enables to decrease time needed for processing and rearrangement of salient points for one and a half. Along with special segmentation and analyses of content, Voronoi diagrams can be used for temporal segmentation of videos, i.e. shot boundary detection. They are also applicable for searching of duplicates and unique frames in video.
video content analyses, segmentation, feature extraction, salient point, Voronoi diagram