Welcome, Guest! Sign Up RSS

Edu Profile

Monday, 06.26.2017
REAL TIME BIOMETRIC FACE RECOGNITION SYSTEM
 [Applied Research : no foreign algorithms or dlls have been used] 

                                               In Fulfillment of

                          Informatics Engineering Junior Project

                                                         By:

                             Ismail Ibrahim Al-Jubbah [Team Leader]

                                  Wassem Ahmad Kassoumeh

                              Muhammad Yasser Hisham Toubaji

                     Ahmad Muhammad Kheir Baney almarjeh

                                               March  2010

Supervised By : Dr. Mohammad Bassam Kurdy

Bookmark and Share     
 
slides link

ABSTRACT:

        Real Time Biometric Authentication using Face Recognition has been a challenging problem for computer vision area. Many previous works have the capability of verifying human faces, but the main task in this research is not only a computer vision system with Face Recognition capability it is also to provide a real-time Biometric authentication with high accuracy using obtained information from the moving object in the scene.

        Our Face Detection algorithm is implemented using the information of the moving object in the scene. The algorithm identification ability gained by analyzing the moving object and depends on extracting the contour of the moving object then comparing it with previously saved samples.

            In order to overcome the statistical based methods problems and to achieve biometric recognition, we depend on building a facial mesh between the eyes, nose, mouth, eyebrows,and extracting the shape of each. Then this mesh will be used later in the recognition process. Face area is passed through a number of preprocessing steps such as noise reduction, scale process and unifying the illumination of the face to obtain better results during the eye, nose, mouth and eyebrow search process.

         

REAL TIME BIOMETRIC FACE RECOGNITION SYSTEM

The main core of our system is the AIU.CV Engine:

[AIU.CV Engine] : Arab International University Computer Vision Engine

[AIU.CV Engine] provide a reusable software components, enable simplified, rapid development of computer vision application. 

This CV engine provide a flexible and reusable software platform which provides all the core functionality needed, right out of the box, to develop a computer vision application while reducing costs, complexities, and time-to-market—all critical factors in the highly competitive CV industry.

[AIU.CV Engine] is a (cross-platform) computer vision framework for the fast and robust tracking and recognizing movable objects in a real-time video stream. [AIU.CV Engine] was mainly designed for the rapid development of camera-based user interfaces.

-  The points behind building our own CV engine  [AIU.CV Engine]:

1-CV Engines which provide a real time response as a camera-based user interfaces with high accuracy and enhanced recognition algorithm are not available for commercial use, and usually CV companies develop their own CV engines for specific environment suitable with their products.

2- And according to our system environment, we can get benefit of extracting movable object from the scene, which increases the accuracy and the performance in our engine.

3- And while our engine should provide biometric face recognition, foreign libraries such like OpenCV and EmguCV become nearly useless while they are based on statistical algorithms to detect and recognize faces, which reduces the accuracy and fail on real time applications.

4- In addition to the previous reason, building our own CV engine gives us the ability to modify or improve the system to make it suitable with all other environments and even to reuse it in future projects.

5-According to our implementation, our infant engine [AIU.CV Engine] has the ability to improve so that it will be able to recognize 3D objects in future work in real time.

 REAL TIME BIOMETRIC FACE RECOGNITION SYSTEM

-Our System limitation is 2D Face Recognition

but, it has the capability of  3D Face Detection

based on the human head contour :


-In order to increase the recognition ratio

we apply Hair Stem Extraction:


-Also we detect the eyes, nose, mouth and eyebrows,

then we extract it's shapes and positions.

The Final Distinctive Mesh:


- Some of our project results:

    

     

     

REAL TIME BIOMETRIC FACE RECOGNITION SYSTEM

- The Hardware of Our Project:

                                

REAL TIME BIOMETRIC FACE RECOGNITION SYSTEM

-The User Interface:

                          

                         

REAL TIME BIOMETRIC FACE RECOGNITION SYSTEM

-The Architecture Diagram: 


REAL TIME BIOMETRIC FACE RECOGNITION SYSTEM

REAL TIME BIOMETRIC FACE RECOGNITION SYSTEM

REAL TIME BIOMETRIC FACE RECOGNITION SYSTEM

REAL TIME BIOMETRIC FACE RECOGNITION SYSTEM