HUMAN MOTION ANALYSIS IN VIDEO SURVEILLANCE SYSTEM

• Main Author: LO , TIMOTHY YIN HONG
• Format: Final Year Project
• Language: English
• Published: IRC 2019
• Online Access: http://utpedia.utp.edu.my/20202/1/Dissertation%20FYP%20II_TIMOTHY%20LO%20YIN%20HONG_23201.pdf; http://utpedia.utp.edu.my/20202/

With the advancement of camera sensor in worldwide, video surveillance is making its way to be a part of human life. Therefore, the demand for an intelligent video analysis has increased. A video sequence consists of sequences of frame, with the detection algorithm, these frames can be analyzed and detect any” “abnormal behavior such as crime.” The essential component of any intelligent video surveillance was detection and tracking, followed by analysis and recognition that classify into type of human motion activities. Besides, the common challenges including occlusion, number of targets, brightness of surrounding and” noise will contribute to some error. GUIDE (Graphic User Interface Development Environment) is provided by Matlab to create a graphic user interface, which ease the analysis of human motion detection, tracking and recognition. In Human motion analysis, the algorithm that used for detection is Histogram of Gradient, followed by Optical Flow coding for a better demonstration of the direction and magnitude of the flow which indicated by a moving object. The direction and magnitude are identified by comparing the current frame to the previous frame, which result in the moving object being detected. In the overview of any kind of video content analysis, the process is generally the same, ranging from input, pre-processing, feature extraction, detection and tracking, classification and recognition and output. A detailed subsystem is introduced and explained in the following Chapter 2 and also discussed in Chapter 4. Aside from the challenge including occlusion and brightness, the number of people to be analyzed and detected in a video frame also can affect the detection rate. The effect of video quality on human detection rate is discussed in Chapter 4. In a human detection, the software is represented by algorithms, in which algorithm also being used in the process of training the dataset. On the other hand, the hardware is represented by the camera that used to capture the video footage that will determine the video footage quality. A good camera can produce high definition videos which further improve the video content analytics and detection rate.