RESEARCH ARTICLE
XXXXXXXXXXXXXXXXX
Copyright © 2016 American Scientific Publishers All rights reserved Printed in the United States of America
Advanced Science Letters Vol. XXXXXXXXX
Android Based Face Recognition System Using PCA and Eigen Face I Nyoman Gede Arya Astawa1, I Gusti Agung Made Sunaya2 , I Gusti Ngurah Bagus Catur Bawa3 123
Department of Electrical Engineering, Bali State Polytechnic - Kampus Bukit Jimbaran, 80714, Bali - Indonesia
Face recognition is one of pattern recognition approachment for personal identification needs beside the other biometrics approachment such as fingerprint recognition, handwriting sign, eye recognition, etc. It is very common on these days for the people to have a mobile phone with integrated digital camera. This provides a good opportunity to develop face recogtion system through the mobile phone. On this paper, android based face recognition system has developed. Approachment method that to be used are PCA (Principal Component Analysis) and Eigen Face. System testing will be done to see how fast the capable of mobile phone to proccess the sistem. This paper shows the mobile phone performance for android based face recognition system. Keywords: PCA, Eigen Face, Face Recognition System, Android.
1. INTRODUCTION Face recognition is one of pattern recognition approachment for personal identification needs beside the other biometrics approachment such as fingerprint recognition, handwriting sign, eye recognition, etc. 1. Face recognition system has been widely used in IT. One of the advantage of face recognition based security system is the security that relatively difficult to be cracked or hacked. Face recognition associated with the object that never the same. This changes caused by face expression, light intensity, angle, or face accessory's change. On these days, it is very common for the people to have a mobile phone with integrated digital camera 2-3. In recent years, more and more people paying attention to the face recognition system in mobile phone 4. This provides a good opportunity to develop face recogtion system through the mobile phone. Moreover, through the bigger internal memory that integrated, it is possible to create the complete training dataset with different lighting, background, pose, expression, etc. 3. . *
Email Address:
[email protected]
1
Most mobile phone has been integrated with one or more camera that possible to create complete training dataset with different lighting, background, pose, expression, etc. 3. Face recognition method based on subspaces analysis such as PCA and LDA, 5 has been improved and started to use in mobile phone. A face verification system is fully automatic in mobile phone was proposed 1, using cascade asymetric PCA (C-APCDA) algorithm.
The development these days has produce so many applications that is using Face's image as information resource. Generally, a face image can give a special information that related to personal identification based on face recognition that can be used on an elctronic security system. To determine the ability of android in terms of accuracy and speed of face recognition, we conduct research using PCA and Eigen Face One of early step in face recognition process is face detection. The process of face recognition system divided into few sub-process. The most important sub-process is face detection 6. On an android based mobile device, face detection using PCA method already included in openCV library. Face Recognition that used is Eigen Face method that included in openCV library too.
RESEARCH ARTICLE
Adv. Sci. Lett. X, XXX–XXX, 2016
2. LITERATURE REVIEW We can look at the researched that 7 have been done, there are various methods and algorithms for the face recognition system. Otherwise, most of them are like Neural Network which mean they work only for single image 8. In the recent 10 years, there are so many algorithms have been developed. 9 did a research related to subpattern based PCA on a face recognition system. they were using PCA method by dividing the pattern from a whole image into subpatterns so that can increase the accuration on a face recognition system. Accuration that obtained from the approach is 95.45%. This value is bigger than PCA without the approach thatis 93.05%. At the same year, 10 developed a same system yet with different approachment that is Twodimensional PCA (2DPCA). Accuration value that obtained is 96.0% from the ORC database and 85.24% from Yale's database. Another approachment related PCA has done by 11. They were using modular PCA approachment. their research shown a problem related to inaccuracies from PCA that caused by various facial expression. In mPCA approachment, image divided into few parts and then PCA applicated into every part of image. The results are not much different from the usual PCA, but on some PCA with various facial expressions, mPCA shown more accurate result. 12 comparing the PCA method with LDA on a face recognition system. They said PCA has given a good result in a various illumination and position. However, PCA has given a great result on a case that has a few training set and huge test dataset. PCA has better accuracy than LDA on facial image that has vaious background. 13 did a face recognition research on mobile device to obtain the image and using PCA method in Matlab for face recognition, where this method was used on media server. The accuration value of the application is 88.88% and the application was used to access control adn perevention of unauthorized use of mobile phone. On this research, the dataset stored in mobile device. 14
developed face recognition system using PCA and Eigen Face method. They said, there are so many method that can be use for biometric system yet face recognition system has the best performance among it. And so the research that has done by 6, their research has used PCA and Eigen Face method for a face recognition system that has been developed. The obtained accuration value of those combination is 92.30%.
Image has taken from the mobile phone camera then do the detection process. Using the PCA that has included in openCV library, system will look into face pattern in the taken image. Cropping be done after the face in image has detected.
Fig. 1. General flow of face recognition system Face recognition process will be done after the cropping process. On recognition process, the image that has been cropped calculated to get the eigen vector and eigen value. Those value will compared with the database. if the taken image matched with the image in the database, process ended. 4. EXPERIMENTAL RESULT A. Dataset
Images that used in this research are 125 facial image from 25 person with 5 variant of pose (fig. 2). Every person's face has taken with 5 variant pose that is facing forward, left, right, up, and down with less than 300 angle. 15.
All those researches above shown the obtained accuration value of PCA and Eigen Face method on a face recognition system is more than 80%. This can be a fondation of PCA and Eigen Face method use that build. 3. RESEARCH METHOD Android based face recognition system using PCA and Eigen Face method figured as Figure 1. In Figure 1, 2
RESEARCH ARTICLE
XXXXXXXXXXXXXXXXX
(a)
(b)
(c)
(d)
Fig. 4. Result attendance : (a) selection attendance (b) selection course (c) input (d) result
Fig. 2. The face image with 5 variant of pose. B. Face detection
We develop an application for android device using Java language. The development has done by integrating native code such as C/C+++ so that will be easy when developing an application that support openCV library. Android Face Detector Application programming Interface (API) tools usually called Face API is the library that used to find face pattern in an Bitmap image. Face image detected by the camera is in the form of RGB image. When the system has detected a face, the face will be marked with a green square (fig. 3). This API included in android.media. FaceDetector that worked by access the findFaces method. Bitmap is representation of image graphic which consist set of dot that stored in memory. The obtained image will converted into grayscale before the detection process. This research is using Android ver 4.0 or API 14.
Table.1. Average time and system accuracy. Average Index of Test Time (sec) Accuracy 1 0.02656 0.88 2 0.02704 0.84 3 0.02672 0.72 4 0.02724 0.84 5 0.02736 0.88 0.026984 0.832 Average Total In Table 1, average total of face recognition system is 0.0269 sec and the accuracy is 0.832 on the range start from 0 to 1. If it converted into precentage so the accuracy is 83.20%. 5. CONCLUSIONS Face recognition system using PCA and Eigen Face method can be implemented in android based mobile device. Using 125 facial image dataset and 5 times testing, the value of average face recognition process time is 0.269 sec and 83.20% of accuracy. This research can improved the accuration value by adding more methiod such as artificial intelligent method. face detection can be improved so in future it can detect multi-face. ACKNOWLEDGMENTS
(a)
(b)
(c)
Fig. 3. Input image face on system. (a) list name (b) image face (c) face detection C. Test result
System test has done 5 times for every person. The process time has been noted and so the accuration value (fig. 4). Table 1 shown the average time and accuracy for every test that has been done.
This work was supported in part by DIPA Unggulan Bali State Polytechnic under Grant No. 04.4895/PL8/LT/2016.
REFERENCE [1]
[2]
3
Wijaya, S.A. Perbandingan Metode Pengenalan Wajah Secara Real-Time Pada Perangkat Bergerak Berbasis Android. Prosiding KNSI 2014 STMIK Dipanegara Makasar, (2014). Chen, B., J. Shen, and H. Sun. A Fast Face Recognition System on Mobile Phone. IEEE International Conference on Systems and Informatics (ICSAI 2012), (2012): p. 1783-1786.
RESEARCH ARTICLE
Adv. Sci. Lett. X, XXX–XXX, 2016 [3]
[4]
[5]
[6]
[7]
[8]
[9]
Peng, P. and Y. Shen. Efficient Face Verification in Mobile Environment Using Component-based PCA. IEEE 6th International Congress on Image and Signal Processing (CISP 2013) (2013): p. 753-757. Li, H. and X. Zhu. of Conference. "Face recognition technology research and implementation based on mobile phone system". in 2016 12th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery (ICNC-FSKD) (2016). Shakhnarovich, G. and B. Moghaddam, Face recognition in subspaces.: Handbook of Face Recognition. Springer London. (2011). Patil, A.M., S.R. Kolhe, and P.M. Patil. Face Recognition by PCA Technique. Second International Conference on Emerging Trends in Engineering & Technology( ICETET), (2009): p. 192-195. Balcoh, N.K., et al. Algorithm for Efficient Attendance Management: Face Recognition based Approach. IJCSI International Journal of Computer Science Issues 9.(4) (2012). Heydarzadeh, Y. and A.T. Haghighat. An Efficient Face Detection Method using AdaBoost and Facial Parts. International Journal of Simulation Systems, Science & Technology (IJSSST). 12.(4) (2011): p. 1-9. Chen, S. and Y. Zhu. 'Subpattern-based Principle Component Analysis. Elsevier - Pattern Recognition. 37.(5)
[10]
[11]
[12]
[13]
[14]
[15]
(2004): p. 1081-1083. Yang, J., et al. Two Dimensional PCA: A New Approach to Appeareance-based Face Representation and Recognition. IEEE Trans. Pattern Analysis and Machine Intelligent. 25.(1) (2004): p. 131 – 137. Gottumukkal, R. and V. K.Asari. An Improved Face Recognition technique based on modular PCA approach. Elsevier - Pattern Recognition. 25.(2004): p. 429–436. Ravi, S., D.P. Mankame, and S. Nayeem. of Conference. "Face recognition using PCA and LDA: Analysis and comparison". in Fifth International Conference on Advances in Recent Technologies in Communication and Computing (ARTCom 2013). Kremic, E., A. Subasi, and K. Hajdarevi. Face Recognition Implementation for Client Server Mobile Application using PCA. Proceeding of the ITI 2012 34th, Int. Conf. on Information Technology Interfaces, Cavta, Croatia, (2012): p. 435-440. Wagh, P., et al. Attendance System based on Face Recognition using Eigen face and PCA Algorithms. IEEE International Conference on Green Computing and Internet of Things (ICGCloT) (2015): p. 303-308. Developers, G. Face Detection Concepts Overview. (2016 [cited 2016 11 Juli].
4
