Real time target identification for security applications

Salex,, Kuan Thai Jun. (2009) Real time target identification for security applications. [Final Year Project Report] (Unpublished)

Preview	PDF REAL TIME TARGET IDENTIFICATION FOR SECURITY APPLICATIONS(24 pgs).pdf Download (132kB) | Preview
	PDF (Please get the password by email to repository@unimas.my , or call ext: 3914 / 3942 / 3933) SALEX KUAN.pdf Restricted to Registered users only Download (1MB)

Abstract

Nowadays, there are a lot of security alarm systems in the markets and those security alarm systems are similar, which is either CCTV video record with alarm siren or security password system with siren, or the combina tion of both. Thus, the real time target identification project for the security appl ications is proposed. In this project, a real time face detection and face recognition sys tem for color image sequences is presented. Basically, the system applies edge detec tion technique for the face localization. The system localizes the human head through outline analysis of boundaries image, then focusing the attention on a specific area in the image. The system applies the same technique to the face recog nition, with combination of neural networks. The system extracts the human fac ial from the color sequences images and simulates the human facial by using the neural networks. The simulation is based on the RGB color layer of images. Neural ne tworks are used to build the database of the system for storing the facial image s and the training data of target facial. The Rprop technique is used for the trainin g process for the system. The system is able to detect the human face and able to recognize the face. The system would show the results of target been identified wi th percentages of matching and together with the image of the target. The system can perform well in recognizing the target and the percentage of matching is above 90%. Besides, the system also able to recognize the target when the target having some changes, example like target wearing hat, veil and etc. The proposed approach i s using a direct input webcam video in real time based for the face detection and face recognition.

Actions (For repository members only: login required)

View Item