In modern cities, massive population causes problems, like congestion, accident, violence and crime everywhere. Video surveillance system such as closed-circuit television cameras is widely used by security guards to monitor human behaviors and activities to manage, direct, or protect people. With the quantity and prolonged duration of the recorded videos, it requires a huge amount of human resources to examine these video recordings and keep track of activities and events. In recent years, new techniques in computer vision field reduce the barrier of entry, allowing developers to experiment more with intelligent surveillance video system. Different from previous research, this dissertation does not address any algorithm design concerns related to object detection or object tracking. This study will put efforts on the technological side and executing methodologies in data visualization to find the model of detecting anomalies. It would like to provide an understanding of how to detect the behavior of the pedestrians in the video and find out anomalies or abnormal cases by using techniques of data visualization.