Start Submission Become a Reviewer

Reading: Forensic art: 2D facial composite through image processing techniques in Sri Lanka

Download

A- A+
dyslexia friendly

Articles

Forensic art: 2D facial composite through image processing techniques in Sri Lanka

Authors:

L. Sivaneasharajah,

University of Colombo School of Computing, LK
X close

M. A. S. Perera,

University of Colombo School of Computing, LK
X close

P. B. Jayasekara,

University of Colombo School of Computing, LK
X close

J. Perera ,

University of Colombo, LK
About J.
Department of Forensic Medicine and Toxicology, Faculty of Medicine
X close

D. D. Karunaratne,

University of Colombo School of Computing, LK
X close

K. D. Sandaruwan,

University of Colombo, LK
About K. D.
School of Computing
X close

R. N. Rajapakse

University of Colombo School of Computing, LK
X close

Abstract

INTRODUCTION

Forensic art is a technique used by law enforcement in order to identify a suspect towards apprehension or conviction of wanted persons and composite art is one branch.  A “facial composite” is a graphical representation of a human face according to eyewitnesses' description.  The failure rate of suspect identification through manual facial composite sketch was reportedly quite high in 2014 in Sri Lanka1.  In addition, international software available for face identification may not be the best approach for Sri Lankan population.

 

OBJECTIVE

To develop an automated, image processing based computer software solution using 2D facial feature templates for suspect identification in Sri Lanka.

 

STUDY DESIGN

Facial feature templates were constructed by analyzing a dataset from Sri Lankan young adult population focusing on most occurring facial indices and commonly available facial feature shapes. Anthropometric proportion indices measurements and shape classification for eyes, nose, face, upper vermilion and lower vermilion of 140 undergraduates of both sexes between 20-25 age were identified and categorized according to BMI (Body Mass Index).  Face Software Development Kit library (Face SDK library) was used for index measurements and domain expert knowledge from aesthetic and forensic art fields were utilized in identifying facial feature shapes. These parameters were incorporated to construct 2D facial feature templates for major facial feature. Then these templates were transferred into the system to be used for composite construction purpose. Finally face visualization process will perform iteratively until the eyewitness is satisfied with the ultimate image. Evaluation of this solution with face pool technique and anthropometric index evaluation techniques obtained 70.19% accuracy and average of 84% respectively.

 

CONCLUSION

A significant accuracy level in composite images has been obtained by using anthropometric indices measurements. Hence, this computerized solution will significantly enhance the currently used manual facial composite procedure for suspect identification in Sri Lanka ensuring justice to the victim of crime.
How to Cite: Sivaneasharajah, L. et al., (2017). Forensic art: 2D facial composite through image processing techniques in Sri Lanka. Sri Lanka Journal of Forensic Medicine, Science & Law. 7(2), pp.24–27. DOI: http://doi.org/10.4038/sljfmsl.v7i2.7779
Published on 03 Feb 2017.
Peer Reviewed

Downloads

  • PDF (EN)

    comments powered by Disqus