The problem of identifying a person by taking an input face image and matching it with a known person in a database has been studied during the 6st semester in 2011 at Aalborg University in Denmark. For this purpose, a 2D and 3D data obtained with the Microsoft Kinect face recognition system is developed. Indeed, widely used methods of 2D and 3D Face Recognition such as Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA) and the Fishersurfaces are analyzed and implemented in C++ with the help of Intel’s Open source Computer Vision (OpenCV). The ICP and the histograms methods are also treated. The Texas 3D Face Recognition Database has been used to test the performances of different combination of algorithms in order to find the most efficient one. A face recognition database has been created using the Kinect as the 2D and 3D sensor then used for a real time face recognition system.

The team

The team was composed by: * **Maxime Biette** * **Thierry Lavoix** * **Brian Lemoine** * **Mickaël Sahnoun** And supervized by **Zheng-Hua Tan**

Files

The source code of the software we used to test our algorithms and the various tools we developed are available on GitHub: https://github.com/BietteMaxime/2D3D-face-recognition-tests It requires both OpenCV 2.2 and Qt4 that you’ll have to get and include separatly.

You can also get the report we wrote in pdf here with a special front page [here)(/files/AAU-FR/FrontPage.pdf).

Go back to the home page