Tracy N. Tacuban (DIT)
Iloilo Science and Technology University, Burgos St. Lapaz Iloilo City,
Philippines
tracy.tacuban@isatu.edu.ph
Date Received: October 17, 2019; Date Revised: April 27, 2020
Asia Pacific Journal of Multidisciplinary Research
Vol. 8 No.2, 158-166
May 2020
P-ISSN 2350-7756
E-ISSN 2350-8442
www.apjmr.com
ASEAN Citation Index
Facial Recognition Using Segmented Facial Points Algorithm for Intelligent Surveillance System 613 KB 2 downloads
Tracy N. Tacuban (DIT) Iloilo Science and Technology University, Burgos St. Lapaz...
In this paper, the researcher presents the Facial Points Segmentation Algorithm as an improved feature matching process of Scale Invariant Feature Transform (SIFT). Because the feature matching process in SIFT would include the features of the image background, errors are automatically
introduced in the matching process. In this study, instead of matching the overall features of an image, the algorithm logically divides the human face into two segments – the eye and face segment. Features from query and gallery/training images are matched only if they belong to the same segment and the similarity of the features is determined using cosine distance. The method is applied in Video Surveillance System designed for automatic face recognition using the profile of people contained therein. Based on the results of the recognition process, the precision rate of the system using F-Measure is 82.25% and
considered “Good” while the functionality of the surveillance system using ISO 25010 metrics is rated Effective” based on the evaluation of ICT professionals and expected users. The importance of the study resides in its potential application in surveillance, attendance and other related systems requiring facial
recognition. Its scheduling system allows the user to set specific time and day to record surveillance video. Meanwhile, its notification system sends email and SMS message to named recipients for intrusion within its confine thereby improving the features of a surveillance system and boosts the sense of security
of the user. The algorithm is terse enough to filter and compare image features within the same face segments but good enough to reduce the processing time as well as limit the possibility of matching error which is necessary for surveillance systems and specifically, for face recognition algorithms which could
be further enhanced by future researchers.
Keywords: Facial Recognition, Segmented Facial Points, Algorithm, Surveillance System, and Biometric Technology