Đang chuẩn bị liên kết để tải về tài liệu:
Mpeg 7 audio and beyond audio content indexing and retrieval phần 10

Trung bình, MPEG-7 ASP dựa trên ICA mang lại hiệu suất tốt hơn so với ASP dựa trên PCA. Tuy nhiên, tỷ lệ công nhận sử dụng MPEG-7 ASP kết quả xuất hiện thấp hơn đáng kể so với tỷ lệ nhận dạng của MFCC. Nhìn chung MFCC đạt được tỷ lệ nhận dạng tốt nhất. | 7.4 HIGHLIGHTS EXTRACTION FOR SPORT PROGRAMMES 259 Table 7.5 Sound classification accuracy FD Feature extraction Holiday Zoo Street Kindergarten Movie Party Average 7 PCA-ASP 92.5 95.5 92.1 91.3 96.5 75.1 90.05 ICA-ASP 91.3 96.2 90.7 90.5 96.9 82.3 91.32 MFCC 97.08 97.6 95.3 96.3 97.6 94 96.31 13 PCA-ASP 96.3 97.6 95.7 95.8 98 82.4 94.3 ICA-ASP 97.9 94.3 96.6 96.6 98.7 93.9 96.33 MFCC 100 99 96.6 99 100 90.1 97.45 23 PCA-ASP 100 98.8 98.5 98.5 100 88.2 97.33 ICA-ASP 99 99.4 97.8 99 100 94 98.2 MFCC 100 100 99 100 100 93.4 98.73 Average 97.12 97.6 95.81 96.28 98.63 88.15 95.56 FD feature dimension. On average MPEG-7 ASP based on ICA yields better performance than ASP based on PCA. However the recognition rates using MPEG-7 ASP results appear to be significantly lower than the recognition rate of MFCC. Overall MFCC achieves the best recognition rate. 7.4 HIGHLIGHTS EXTRACTION FOR SPORT PROGRAMMES USING AUDIO EVENT DETECTION Research on the automatic detection and recognition of events in sport video data has attracted much attention in recent years. Soccer video analysis and events highlights extraction are probably the most popular topics in this research area. Based on goal detection it is possible to provide viewers with a summary of a game. Audio content plays an important role in detecting highlights for various types of sports because often events can be detected easily by audio content. There has been much work on integrating visual and audio information to generate highlights automatically for sports programmes. Chen etal. 2003 described a shot-based multi-modal multimedia data mining framework for the detection of soccer shots at goal. Multiple cues from different modalities including audio and visual features are fully exploited and used to capture the semantic structure of soccer goal events. Wang etal. 2004 introduced a method to detect and recognize soccer highlights using HMMs. HMM classifiers can automatically find temporal changes of events. In this .