A novel highlight event decision approach for baseball videos

Abstract

A real-time highlight extraction system using the caption information has been proposed to detect and classify the highlight events of the baseball games. The system contains several stages: caption extraction, caption identification, content recognition, and model-indexing decision stages. A superimposed caption in the baseball videos is extracted using a multi-frame averaging technique. After extracting the caption, the caption is identified into one of the caption types. The corresponding caption model including the size, position and meaning of each caption content is used to segment the content of the caption accurately. According to the information of the change of the caption content, a novel model-indexing decision approach is proposed to detect and classify the highlight events of the baseball videos. The experimental results show that the proposed approach is very efficient to classify the highlight events. Furthermore, the performance of the proposed approach can be improved with 14% over the rule-based approach.