论文信息 - SFINX: a multisensor fusion and mining system

SFINX: a multisensor fusion and mining system

In a surveillance system, video signals are generated by multiple cameras with or without spatially and temporally overlapping coverage. These signals need to be transmitted, processed, fused, stored, indexed, and then summarized as semantic events to allow efficient and effective querying and mining. SfinX aims to build several core components for multisensor fusion and mining. This paper first depicts SfinX' architecture and its core components, namely, data fusion, event detection, event characterization, event recognition, and storage. In particular, we survey representative methods and discuss plausible research directions for event recognition and storage.

Zoran Dimitrijevic | G. Wu | E. Y. Chang

[1] Edward Y. Chang,et al. The XTREAM multimedia system , 2002, Proceedings. IEEE International Conference on Multimedia and Expo.

[2] Edward Y. Chang,et al. Class-Boundary Alignment for Imbalanced Dataset Learning , 2003 .

[3] Trevor J. Hastie,et al. Discriminative vs Informative Learning , 1997, KDD.

[4] Edward Y. Chang,et al. BubbleUp: low latency fast-scan for media servers , 1997, MULTIMEDIA '97.

[5] Takeo Kanade,et al. A System for Video Surveillance and Monitoring , 2000 .

[6] Edward Y. Chang,et al. Design and Implementation of Semi-preemptible IO , 2003, FAST.

[7] Edward Y. Chang,et al. Adaptive Feature-Space Conformal Transformation for Imbalanced-Data Learning , 2003, ICML.

[8] Christopher J. C. Burges,et al. Geometry and invariance in kernel based methods , 1999 .

[9] Si Wu,et al. Improving support vector machine classifiers by modifying kernel functions , 1999, Neural Networks.