UAD Lecturers' introductory system through surveillance cameras with eigenface method

Abstract

Technologies related to processing using computers are developing so rapidly, such as applications to identify a person automatically through camera monitors (CCTV). The human recognition application in real time can be found in the surveillance system, identification and facial recognition. The direct observation of human beings has a weakness such as fatigue and saturation that may occur, resulting in decreased accuracy. For that, computer can be an alternative solution to overcome it. For example, the human Face Recognition (Eigenface) detection system. This system can be very helpful when you want to find and know the existence of someone in a place, for example to help in finding the existence of lecturers on campus. Students often seek lecturers to conduct guidance or for other academic matters, but students often do not know whether the lecturers sought on campus or not. Therefore, in this research an application will be made to help students in knowing the existence of lecturers on campus. This final project examines the system to recognize lecturers who are on campus using CCTV. The method used is eigenface. Eigenface is one of the facial pattern recognition algorithms based on the Principle Component Analysis (PCA). The basic principle of facial recognition is to cite the unique information of the face and then be encoded and compared with the previously done decode result. The process itself consists of data collection and facial recognition processes. In the process of collecting data, the data taken in the form of the name and the image of the lecturer will be used as a database to recognize the face of the lecturer. While the facial recognition process is the process by which the face of the lecturer who has been caught by the camera will be compared with the database that has been taken to recognize the lecturer. From the research done can be concluded that there are several factors that affect the accuracy of the system including the distance of the camera sensor with the most effective object is 1 to 2 meters, the intensity of bright or dim light, the face positioning and Number of datasets owned. The test results obtained an accuracy of 89%.