Algorithms and System for High-Level Structure Analysis and Event Detection in Soccer Video
Peng Xu, Shih‐Fu Chang, Ajay Divakaran, Anthony Vetro, Huifang Sun · 2001
In this report, we first present a general framework for video structure and content analysis. In this framework, frame-based low-level features are extracted. Each frame is represented by the values of features or labels converted from the features. So video sequence is transformed into multiple label sequences or real number sequences. Each of such sequence is associated with one of the extracted low-level feature. The feature sequences are analyzed together to extract high-level semantic features. Based on this framework, we describe an application system specifically for soccer video indexing and summarization. We use a distinctive feature to capture the high-level structure of the soccer video (e.g., play boundaries) and use a unique feature, grass orientation, together with camera motion to detect interesting events such as play strategy. The unique features of the system include compressed-domain feature extraction for real-time performance, use of domain specific features for detecting high-level events, and integration of multiple features for content understanding.