Object detection and tracking motion for event analysis

• Main Author: Tan, Nicholas Sum Jun.
• Other Authors: Chan Kap Luk
• Format: Final Year Project
• Language: English
• Published: 2012
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Control and instrumentation
• Online Access: http://hdl.handle.net/10356/49895
• Institution: Nanyang Technological University

This project titled “Object Detection and Tracking Motion for Event Analysis” focuses on some computer vision techniques that detects and tracks objects that are in motion in a video. The author’s task is to learn a few detection and tracking methods and attempt to track a video sequence. To detect any objects in motion, one of the background subtraction methods, Mixture of Gaussians, is implemented. In order to track multiple objects across frames, one of the minimum cost flow algorithm Successive Shortest Path algorithm is implemented to associate the location data of each object across the frames in order to obtain the object’s track. After obtaining the tracks, all information of the tracks such as X-Y position, size of object, minor/major axis length and frame number is written into Excel file to form a database. A classification of the detected objects is also attempted by using the values of minor and major axis length to differentiate pedestrians and vehicles.