ADAPTIVE ALGORITHMS OF FACE DETECTION AND EFFECTIVENESS ASSESSMENT OF THEIR USE

Article Sidebar

PDF
DOI: https://doi.org/10.20998/2522-9052.2018.3.02
Keywords:
face detection and recognition, the Viola-Jones method, an automatic stabilization of frame brightness, the probability of face and its main elements correct detection

Main Article Content

Kostiantyn Dergachov
National Aerospace University – Kharkiv Aviation Institute, Kharkiv
https://orcid.org/0000-0002-6939-3100
Leonid Krasnov
National Aerospace University – Kharkiv Aviation Institute, Kharkiv
https://orcid.org/0000-0003-2607-8423
Oleksandr Cheliadin
National Aerospace University – Kharkiv Aviation Institute, Kharkiv
Anatoliy Zymovin
National Aerospace University – Kharkiv Aviation Institute, Kharkiv

Abstract

Subject of research is the detection and recognition of faces. The purpose of this work is creation of modified algorithms of face detection, which are providing automatic brightness stabilization of the analyzed image regardless of brightness level. A technique is proposed for assessing the effectiveness of their work in comparison with the classical algorithm. Research methods. We will dwell in more detail on the first part of the problem − face detection and recognition. In the meantime, the most popular method used for searching the face area on an image is the Viola-Jones method, which is popular because of its known high speed and efficiency. It is based on an integral image representation, on the method of constructing classifiers based on adaptive boosting algorithm (AdaBoost) and on the combination classifiers in cascade structure method. The Viola-Jones method is firstly using cascades of wavelets (primitives) - Haar features. All of the above made it possible to build a face detector that works in real-time with a fairly high quality. However, there are a lot of disturbing factors, which are limiting the efficiency of such algorithm work. The major of them are spacial face position ambiguity on the analised image and poor quality of stage lighting. The results of the study. The adaptive algorithms of face detection and recognition on digital images and video sequences in real-time, based on the Viola-Jones method, are suggested. An automatic stabilization of frame brightness is additionally added to the classical structure of such algorithms to compensate an effect of changes in the stage illumination level on quality of face detection. The structure of the algorithms is described and the software developed in Python programming language for a face detection and recognition using OpenCV library resources. Video data is processed in real time. An original method for the efficient estimating of the algorithm based on the criterion of the maximum probability of faces and their main elements (eye, nose, mouth) correct detection is proposed and implemented programmatically. The results of work of classic and suggested algorithms are compared. The examples of work and testing of software are given. Conclusion. The use of the obtained results allows to improve the quality of work and the reliability of the results when recognizing faces in different systems.

Article Details

How to Cite
Dergachov, K., Krasnov, L., Cheliadin, O., & Zymovin, A. (2018). ADAPTIVE ALGORITHMS OF FACE DETECTION AND EFFECTIVENESS ASSESSMENT OF THEIR USE. Advanced Information Systems, 2(3), 10–18. https://doi.org/10.20998/2522-9052.2018.3.02
Issue
Vol. 2 No. 3 (2018): Advanced Information Systems
Section
Identification problems in information systems
Author Biographies

Kostiantyn Dergachov, National Aerospace University – Kharkiv Aviation Institute, Kharkiv

Candidate of Technical Sciences, Associate Professor, Head of Aircraft Control Systems Department

Leonid Krasnov, National Aerospace University – Kharkiv Aviation Institute, Kharkiv

Candidate of Technical Sciences, Senior Research Fellow, Associate Professor of Aircraft Control Systems Department

Oleksandr Cheliadin, National Aerospace University – Kharkiv Aviation Institute, Kharkiv

Postgraduate student of Aircraft Control Systems Department

Anatoliy Zymovin, National Aerospace University – Kharkiv Aviation Institute, Kharkiv

Candidate of Technical Sciences, Senior Research Fellow, Associate Professor of Aircraft Control Systems Department

References

Viola, P. and Jones, M.J. (2001), “Rapid object detection using a boosted cascade of simple features”, IEEE Conf. on Computer Vision and Pattern Recognition, Kauai, Hawaii, USA, Vol. 1, pp. 511–518.

Viola, P. and Jones, M.J. (2001), “Robust real_time face detection”, International Journal of Computer Vision, Vol. 57, No. 2, pp. 137–154.

Joseph, Howse and Joe, Minichino (2015), Learning OpenCV 3 Computer Vision with Python : Second Edition, Packt Publishing, September, ISBN : 978-1-78528-977-4.

Saurabh, Kapur (2017), Computer Vision with Python 3, Packt Publishing, August 2017, ISBN: 978-1-78829-976-3.

Prateek, Joshi (2015), OpenCV with Python By Example, Packt Publishing, September 2015, ISBN: 978-1-78528-393-2.

Open Source Computer Vision, available at : http://docs.opencv.org/trunk (last accessed March 7, 2018).

Bovyrin, A., Druzhkov, P. and Erukhimov, V. (2016), Development of multimedia applications using libraries OpenCV and IPP, Intuit, Moscow, 382 р.

Comasch, F. Stuijk, S. Basten, T. and Corporaal H. (2013), “Rasw: A runtime adaptive sliding window to improve viola-jones object detection”, Distributed Smart Cameras (ICDSC), 2013 Seventh International Conference.