In this paper, we present an audio-based event detection algorithm shown to be effective when applied to Soccer video. The main benefit of this approach is the ability to recognise patterns that display high levels of crowd response correlated to key events. The soundtrack from a Soccer sequence is first parameterised using Mel-frequency Cepstral coefficients. It is then segmented into homogenous components using a windowing algorithm with a decision process based on Bayesian model selection. This decision process eliminated the need for defining a heuristic set of rules for segmentation. Each audio segment is then labelled using a series of Hidden Markov model (HMM) classifiers, each a representation of one of 6 predefined semantic content classes found in Soccer video. Exciting events are identified as those segments belonging to a crowd cheering class. Experimentation indicated that the algorithm was more effective for classifying crowd response when compared to traditional model-based segmentation and classification techniques. 1

Abstract. In this paper we have proposed a scheme for parsing News video sequences into their semantic components using integrated aural and visual features. We have explored use of the Token Passing Algorithm with HMM for simultaneous segmentation and characterization of the components. Experimentation with about 100 sequences have shown impressive results. 1

This thesis has focused on the automatic video indexing of sports video, and in par-ticular the sub-domain of football. Televised sporting events are now commonplace especially with the arrival of dedicated digital TV channels, and as a consequence of this, large volumes of such data is generated and stored online. The current process for manually annotating video files is a time consuming and laborious task that is essen-tial for the management of large collections, especially when video is often re-used. Therefore, the development of automatic indexing tools would be advantageous for collection management, as well as the generation of a new wave of applications that are reliant on indexed video. Three main objectives were addressed successfully for football video indexing, con-centrating specifically on audio, a rich and low-dimensional information resource proven through experimentation. The first objective was an investigation into football video domain, analysing how prior knowledge can be utilised for automatic indexing. This was achieved through both inspection, and automatic content analysis, by applying the