文章基本信息

标题：Motioned Facial Recognition from Live Feed for Surveillance Solutions
本地全文：下载
作者：Sarah Bal ; Anmol Kalra ; Rishi Kumar 等
期刊名称：International Journal of Soft Computing & Engineering
电子版ISSN：2231-2307
出版年度：2014
卷号：4
期号：2
页码：195-201
出版社：International Journal of Soft Computing & Engineering
摘要：The paper focuses on how face recognition can be done on live video stream (using a webcam-inbuilt or USB attached).The live video is checked for any human face. If a human face is detected, a rectangular box is formed around the face. If nothing is found for the face detection method, a text box showing the error is presented in front of the user. If the face is detected this face is then matched with the already saved database which was priorly created having images of different faces. This is the training database which is then matched with the face image extracted from the live video stream. Initially the project shows the process of face detection and matching procedure from images and then proceeds to face recognition and matching through a live video streaming. The live video here considered is the webcam, the face is detected through the webcam and if any match is found from the train database previously stored in the computer or the device is found then both the detected image and the current image are displayed on the graphical user interface. The GUI being made consists of three axes windows, one showing the continuous live streaming of video, the second shows the screenshot or singular frame of the face detected in the live stream and the third has the image got from the database that somewhat matches to the current image being displayed. The two databases are there, one the train database where the images of different faces of people are stored which would then be used for matching from live video stream for the purpose of security and authentication. The test database consists of the images that are being received from the live video stream, the video stream as soon as it detects the face of human, takes the snapshot of the frame and saves it to the test database, these images are then checked for authentication by matching them with the images in the train database.