Real-time systems for moving objects detection and tracking using pixel difference method.

摘要

Tracking of multiple moving objects from 'real world' in an image sequence is a very important task in Computer Vision. In particular, there are many problems in which objects undergoing motion must be detected and tracked. In this thesis, two tracking systems are introduced: Fish-Tracker System and Human-Counter System. In these two systems, the objects are tracked through a sequence of frames and the motion detection algorithm is used based on frame pixel difference. Fish-Tracker system is developed for department of Biological Sciences to determine spent time, path and velocity of moving fish. The direction and velocity of each object are calculated between each pair of frames and are used to predict the position of the object in the next frame. The calibration, reconstruction procedure and the problems due to the presence of noise and shadows in the images are also addressed. Human-Counter system is kind of surveillance system that is important for the office security or the marketing research. In this system, some basic image morphological operations such as opening and closing are employed to reduce noise. The adopted strategies such as a connected components algorithm and boundary extraction are detailed.