Semantic event detection in sports video using Hidden Markov Model

In this paper we propose semantic event detection in sports video using Hidden Markov Model. It is reasonable to model the events of sports video using Hidden Markov Models because sports video could be considered as a rule-based sequential signal. Our goal is to identify and classify the events of interest to facilitate highlight-based video indexing and summarization. In this paper, we define 3 semantic events of golf spots video, and design HMM model for each event. Furthermore, we use 10 multiple visual features based on MPEG-7 visual descriptors to acquire parameters of HMM. We have tested the proposed scheme on a video database. Experiment results have been shown to provide very reasonable performance for identifying a variety of golf events.