Mining trajectory log for patterns and anomalies

Abstract

The need of detection of patterns and behaviors has been increasing in demand in the recent years as the quantity of moving objects rises. Examples of moving objects can be vehicles, human beings, animals or even vessels. By acquiring the positions of moving objects and analyzing them, we can find out the behaviors of the subjects (moving objects).

Any behavior that deviates from the normal pattern can be used to interpret as urgent or even important to the subject. There are existing sources, reports on the geometric attributes of the positions, trajectories of moving objects; however the other important properties such as the semantics and the background geographical information are often left out.

The objective of this FYP is to design and implement a program to do detection of patterns and moving objects anomalies from historical logs. The program will take in files containing geometric attributes of a human being and converting the data into a file that can be displayed onto Google Earth. Based on the current geometric position of the subject and the historical logs of previous travels, the program can detect any abnormal patterns and behaviors made by the subject.