Detecting Automobiles and People for Semantic Video Retrieval

This paper describes a method for detecting automobiles and people in streaming or archived video. Our video object tracking system is based on Kalman filter updating of an active contour over the video sequence. We use the sequential probability ratio test (SPRT) to classify the moving objects. Results are shown of a real video sequence from a busy city intersection.

• Type: Conference
• Year: 2002
• Where: ICPR
• Authors: Rene Visser, Nicu Sebe, Michael S. Lew
• Keywords: ICPR 2002, Real Video Sequence, Sequential Probability Ratio Test, Video Sequence