[ieee 2014 6th international conference on computer science and information technology (csit) -...

5

Click here to load reader

Upload: adamu

Post on 10-Feb-2017

217 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: [IEEE 2014 6th International Conference on Computer Science and Information Technology (CSIT) - Amman, Jordan (2014.03.26-2014.03.27)] 2014 6th International Conference on Computer

Automatic Face Reconstruction System Nur Nazihah Rahim, Nur Azrin Abd Malek, Akram M. Zeki, Adamu Abubakar

Department of Information Systems International Islamic University Malaysia, Malaysia

[email protected], [email protected], [email protected], [email protected]

Abstract— A face can be sketch by freehand based on the evidence of an eyewitness description which is necessary for the Detectives to tract the looks of the offender. This process is slow; moreover there are some factors which cannot be assumed by the artist, including weight, hair color and length, and eye color. However, if this process is conducted correctly, the result should bear a striking resemblance to the individual. As with composite drawings, an image of the resulting face can then be distributed among law enforcement personnel or to the general public. The reconstructed face should be photographed in such a way that there is no perspective distortion. This paper presents Face Reconstruction for Identification that approach is relatively same with the sketching by freehand. However, this system will reconstruction and generate the face automatically. This will create an effectiveness and efficiency in identifying and recognize the offender

Keywords- Detective, offender, face reconstruction, identification

I. INTRODUCTION The face is one of most important part of human beings to

differentiate person’s identity. In general, the term “face identification” can be defined as an ability to recognize people by their facial characteristics. It would match one face to another, probably taken at different time and different condition. When a person committed a crime, usually witnesses take notice of the offender. In order to find the offender, an artist needs to draw or sketch the face of the offender according to information that is gained from the witness. This is done by asking a witness the question about facial characteristics of the offender. The explanation of the witness is important during the interview. Every effort should be made to help the witness feel as happy, relaxed, and comfortable as possible. The witness's comfort takes precedence over the artist's or any other individual’s comfort during the act of drawing the suspect’s face from the witness's memory [1]. The mood of the artist is also important, and it can be controlled and nurtured from within by understanding that even the most imperfect work can be a factor in solving the case. If in the beginning artist knows how many varied ways a sketch can help the investigator identify the offender, it will give him or her confident attitude [1].

A list is provided to help instill in the artist a positive attitude that will produce the best efforts. This positive attitude is essential for the artist cannot instill confidence in a witness if he doesn’t first feel confident himself. Some scenarios where forensic sketches helped identify the perpetrator are illustrated in [1]. Finally, the question “How long after the crime should the sketch be done?” is addressed [1]. Configure system is always necessary for face recognition, and appears to support

what remains of face identification even in prosopagnosic people who have an intact part-based system [2]. Thus configure and part-based systems are needed in face recognition. The database of facial expression has been build for seven expressions: ‘‘neutral”, ‘‘happy”, ‘‘sad”, ‘‘surprised”, ‘‘angry”, ‘disgusted” and ‘‘bored-sleepy” in this could be an aspect which will might add up to the face recognition system [3]. From the evidences above, a solution to overcome these problems are needed. This plan in creating a system about face reconstruction. After all Automatic Face Reconstruction for Identification (AFRFI) System was proposed. The approach is relatively same with the sketch by freehand reconstruction only that we generate the face automatically.

The objectives of this system are given the Detectives an easy way to detect the offender. The Detective does not have to sketch the offenders but by choosing facial characteristics such as eyebrows, eyes, nose and mouth without freehand sketching. The pictures of facial characteristics are stored in the database which contains details of particular person and the characteristics of their facial features. From the facial feature selection, it will match to the database and the actual result of the offenders will be appearing based on the selection made. The selected pictures of facial characteristics will assemble to show a complete picture of the offender.

II. RELETED WORKS

There are many related research approaches for designing face reconstruction system [4-9]. Crucial to that are Forensic sketch face, Pimp the Face and Digibody's Caricature Maker. Forensic sketch face: This system is about how Detectives sketch a face based on the evidence of an eyewitness. It is very necessary for the police when the eyewitness gives them a rough description as to how the fugitive or the offender looks like. Everyone can be sketching the face even though they are not an artist. This system provides a large number of each facial component including eyes, nose, mouth, hair, head, eyebroess, glasses, moustache, jaw and beard. It also provides size to change the facial part because not everyone will be of the same length and breadth. The transparency level of any feature facial part also is provided. It can do by first clicking on a particular feature and then drag the transparency pointer towards left or right to increase or decrease transparencies.

Pimp the Face: This system gives to users to choose a variety number of facial characteristics such are jaw, beard, head, hair, mouse, moustache, eyes, glasses and nose. The facial part can be scaled to the exact position. It has been also given user to adjust the opacity of facial parts.

2014 6th International Conference on CSIT ISBN:987-1-4799-3999-2

208 978-1-4799-3999-2/14/$31.00©2014 IEEE Published by the IEEE Computer Society

Page 2: [IEEE 2014 6th International Conference on Computer Science and Information Technology (CSIT) - Amman, Jordan (2014.03.26-2014.03.27)] 2014 6th International Conference on Computer

Figure 1: Forensic sketch face [4]

Users can delete or undo the picture selection. After users select all the picture of facial parts, users can make the picture whether clear all, duplicate, save, load the face, snapshot and print the face picture. This system gives user friendly whenever users use this system. They have many options to choose after they had selected the complete picture.

Figure 2 : Pimp the Face [5]

Digibody's Caricature Maker: The caricatures were broken them down into parts which are face, mouth, nose, eyes, eyebroess, ears and hair. The caricature uses those elements to create new and unique caricatures. User can have many selections of facial parts to choose. It has adjust button (up, down, left, right) to move the position of each facial part. User can save the complete picture as image in .jpg, .gif, .png as a file format. It is also can be save as html code and save caricature scrip.

Figure 3: Digibody's Caricature Maker [6]

III. SYSTEM DESIGN

A prototype is design in order to enhance the previous related systems reviewed. Our approach will help in assisting the user to better understand reconstruction system. The

approach is an iterative process and the entire generic software development process is applied to our development.

A. Analysis Requirement Specification The first step in the design, just as it applies to any other

software design is undertaking the requirement analysis. The result of requirement specification yield necessary tools, features and functions that can be included in the system. For our system, the tools, features and functions of our system are presented below:

Software

The operating system used is Window 7, and the database management system (DBMS) used is MySQL in order to store and manage all data of the facial characteristics. Adobe Photoshop CS4 is used for editing pictures and text whereas Adobe Dreamweaver CS 4 is used to write the code. Xampp Control Panel 1.8.1-0 is used as the application server and finally, Paint and Microsoft Office Visio 2007 are used to edit picture and creates scripts respectively.

Hardware

A PC with the following configuration is used: Intel(R) Core(TM) 2 Duo CPU 2GHz. 1GB DDR3 Graphic Card. 500 GB Hard Disk Drive memory and another PC with same configuration is used as the server.

Features and Functions

The features and functions of the system during requirement analysis were resolved to categorize them into two modules; the detective module and admin module.

The admin module is the administrative part of the system, were the entire system is control. The admin is a person responsible for managing every aspect of the system and have full control of the system. The detective modules provide interfaces for reporting cases of any facial attribute witness in order to be constructed.

In the admin module, an interface for logging is provided, where the admin will log in first by entering the username and password. On like the admin module, in the detective module login-in for registration is required first. This means that the detective needs to sign up first by filling up a form that to submit some particular details required. After that, he/she is allowed to log in. When the login page appears, detective needs to enter their email as a username and password to enable them access to the system. The Detective needs to re-enter their username and password they entered incorrectly. After username and password are accepted by the system, success login page will appear. After that, detective can be able to report the facial description by selecting the facial characteristics from the ranges of facial attribute provided by the system. The result of the facial display will be a close to a perfect match of a target person’s picture this can automatically update admin module. Then, the admin can view result and make a decision on whether to keep or delete or view the result.

B. Data Flow Model of the System The data flow model presents the logical flow of information through the system. This is described as the business logic of

Identify applicable sponsor/s here. If no sponsors, delete this text box. (sponsors)

2014 6th International Conference on CSIT ISBN:987-1-4799-3999-2

209

Page 3: [IEEE 2014 6th International Conference on Computer Science and Information Technology (CSIT) - Amman, Jordan (2014.03.26-2014.03.27)] 2014 6th International Conference on Computer

the system. A detailed data model shows the interrelationships among the different parts system (see Figure 4). The data flow is controlled by the administrator as shown in Figure 4. The system will received the data from the detective and the admin proceed it.

Figure 4: Data Flow Diagram Stage.

Having the analyzed the tools features and functions required for the system, the next step is to coding. Our codes were designed and the development is produced from the code base on the functional flow presented in Figure 5.

Figure 5: Functional Flow of the design.

IV. RESULT

The system developed was run and tested. The detectives module is describe as the input to the system, in the sense that detectives will report cases by selecting the facial features provided by the system. The facial features provide are given in Figure 5 through Figure 8.

Figure 5: Eyebrows Page

Figure 5: Eyes Page

2014 6th International Conference on CSIT ISBN:987-1-4799-3999-2

210

Page 4: [IEEE 2014 6th International Conference on Computer Science and Information Technology (CSIT) - Amman, Jordan (2014.03.26-2014.03.27)] 2014 6th International Conference on Computer

Figure 6: Nose Page

Figure 7: Mouth Page

Figure 8: Mouth Page

Figure 9: Reconstructed face page.

Figure 9 is the reconstructed faces form as a result of selection of the facial features from the facial attribute pages. The selection is on the basis of description, where the detective listens to the features of face reported by the witness. As the witness describes the facial feature, the detective selects the close match base on the witness description. When this phase is done, the next is to show the witness the reconstructed face and ask if it matches what the witness sees. If it matches, then the page can be saved, otherwise the detective has to go back from the begin and began to select close matches again.

V. FEATURE EBNHANCEMENT The system can be enhanced by adding more entity that can

involve many facial attribute and management structure. For instance, the administrator can collaborate with the other organizations related to, security in order to monitor and track to a reasonable degree of accuracy the face that witness report.

VI. CONCLUSION

This paper presents the proof of concept on automatic facial reconstruction. The system is aim to fulfilled some objectives that have been set in order automatically generate in reconstructed the facial attribute, similar to the traditional method which is used by freehand sketching. This system is more effective and efficient in order to identify and recognize faces than the freehand sketching, and give accurate result of the image appear compared to the traditional way that give inaccurate results in terms of size, shape and color of facial attribute. The motivation of the work lies within the safety of the humankind as well as the peacefulness of the worldwide. [12-13].

REFERENCES [1] L. Gibson, "Pulling Faces From Witness Memory" Forensic Art

Essentials, 2008, Pages 82-123 [2] J. Rivest, M. Moscovitcha, S. Black, A comparative case study of

face recognition: The contribution of configural and part-based recognition systems, and their interaction, Neuropsychologia 47 (2009) 2798–2811.

2014 6th International Conference on CSIT ISBN:987-1-4799-3999-2

211

Page 5: [IEEE 2014 6th International Conference on Computer Science and Information Technology (CSIT) - Amman, Jordan (2014.03.26-2014.03.27)] 2014 6th International Conference on Computer

[3] O. Stathopoulou, E. Alepis, G.A. Tsihrintzis, M. Virvou, "On assisting a visual-facial affect recognition system with keyboard-stroke pattern information, Knowledge-Based Systems, 23, 4, May 2010, pp. 350-356.

[4] A. Lanitis, C.J Taylor, T.F Cootes, Automatic face identification system using flexible appearance models, Image and Vision Computing, 13(5), 1995, pp.393-401.

[5] H. Liao, Q. Chen, Q. Zhou, L. Guo, Rapid 3D face reconstruction by fusion of SFS and Local Morphable Model, Journal of Visual Communication and Image Representation, 23(6), 2012, pp. 924-931.

[6] D. Jiang, Y. Hu, S. Yan, L. Zhang, H. Zhang, W. Gao, Efficient 3D reconstruction for face recognition, Pattern Recognition, 38(6), 2005, pp. 787-798.

[7] P. Claes, D. Vandermeulen, S. De Greef, G. Willems, J. G. Clement, P. Suetens, Bayesian estimation of optimal craniofacial reconstructions, Forensic Science International, 201(1–3), 2010, pp. 146-152.

[8] X. Xie, K. Lam, Face recognition using elastic local reconstruction based on a single face image, Pattern Recognition, 41(1), 2008, pp. 406-417.

[9] Forensic sketch face. Retrieved October 28 2013. From http://flashface.ctapt.de/

[10] Pimp The face System. Retrieved November 15 2013. From http://www.pimptheface.com/create/

[11] Digibody’s Caricature Maker. Retrieved November 15 2013. From http://digibody.com/avatar-maker/index.php

[12] Surah Al-Ma’idah [5:32] Sahih International - The Holy Qur'an - ريم رآن الك http://quran.com/5/32 - الق

[13] Surah Al-Nahl [16:90] Sahih International - The Holy Qur'an - ريم رآن الك http://quran.com/16/90 - الق

2014 6th International Conference on CSIT ISBN:987-1-4799-3999-2

212