TY - GEN
T1 - Performance Analysis of Parallel Implementation of PCA-based Face Recognition using OpenCL
AU - Sapna, S.
AU - Anjali, R.
AU - Kamath, Shobha N.
PY - 2019/5
Y1 - 2019/5
N2 - Face is one of the most commonly used biometrics for identifying a person uniquely. Face Recognition is a technique of identifying an individual and verifying their identity using the face. It finds applications in many areas including forensic investigations, security systems, mobile applications etc. All these systems require fast face recognition rate. But the complex structure of the face makes it difficult to speed up the process since huge computations are involved. One of the well known algorithms for face recognition is the Principal Component Analysis (PCA) Approach. This technique is very effective except for its higher computational cost. In this paper, PCA based face recognition is parallelized using OpenCL to speed up the computation and a comparison is made in terms of computational time required for serial and parallel implementation. Parallel implementation offers a 32-fold speedup over serial implementation proving the efficiency of the proposed method.
AB - Face is one of the most commonly used biometrics for identifying a person uniquely. Face Recognition is a technique of identifying an individual and verifying their identity using the face. It finds applications in many areas including forensic investigations, security systems, mobile applications etc. All these systems require fast face recognition rate. But the complex structure of the face makes it difficult to speed up the process since huge computations are involved. One of the well known algorithms for face recognition is the Principal Component Analysis (PCA) Approach. This technique is very effective except for its higher computational cost. In this paper, PCA based face recognition is parallelized using OpenCL to speed up the computation and a comparison is made in terms of computational time required for serial and parallel implementation. Parallel implementation offers a 32-fold speedup over serial implementation proving the efficiency of the proposed method.
UR - http://www.scopus.com/inward/record.url?scp=85082394663&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85082394663&partnerID=8YFLogxK
U2 - 10.1109/RTEICT46194.2019.9016732
DO - 10.1109/RTEICT46194.2019.9016732
M3 - Conference contribution
AN - SCOPUS:85082394663
T3 - 2019 4th IEEE International Conference on Recent Trends on Electronics, Information, Communication and Technology, RTEICT 2019 - Proceedings
SP - 877
EP - 881
BT - 2019 4th IEEE International Conference on Recent Trends on Electronics, Information, Communication and Technology, RTEICT 2019 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 4th IEEE International Conference on Recent Trends on Electronics, Information, Communication and Technology, RTEICT 2019
Y2 - 17 May 2019 through 18 May 2019
ER -