| Author | Suwan Tongphu |
| Note | A thesis submitted in partial fulfillment of the requirements for the
degree of Master of Science in
Computer Science, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Abstract | Behavior detection has recently became a very interesting topic in computer vision and
intelligent system. Many techniques involving this area has been recovered and contin-
uously adapted in order to prevent unexpected events. In this paper we aim to detect
human suspicious activities in a wide open parking area which seem to be harmful to
the person property. Especially their cars which are difficult to be monitored by the
human effort. The algorithm is divided into two main parts consisting of the object
detection capable of detecting multiple objects in the same scene and the behavior
modeling describing the relationship among the moving blob and the object instances.
In the first section, we use frequency vectors to represent the image regions obtaining
by running sliding windows over the image then train the classifier using AdaBoost.
In order to detect abnormal activity we do experiment on multiple techniques using
statistical models (CRFs and HMM) and a rule base system. The results obtaining
from CRFs and rule base system are dramatically effective while HMM is fair in result.
The experiment indicates that a good result of the modeling could be obtained when
we have a good detection rate of the objects of interest. |
| Year | 2009 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Computer Science (CS) |
| Chairperson(s) | Dailey, Matthew N.; |
| Examination Committee(s) | Honda, Kiyoshi;Haddawy, Peter; |