Abstract
Video content processing and analysis is going through a transition from low-level feature based techniques to high-level semantics based approaches. In this paper, we describe such a transition that low level features are processed to extract four semantic patterns, leading to high-level content analysis for snooker videos. Such extracted semantics and recognised patterns include: (i) full court scenes for snooker match, (ii) close-up view of snooker match; (iii) player’s face, and (iv) audience’s faces. Experimental results support that the proposed technique works well with snooker videos, providing a significant potential for automatic snooker video processing such as annotation, summarization and editing.