3D Face recognition related with facial expressions based on MB-LBP method

Résumé

ABSTRACT

3D Face Recognition is a promising alternative to solving problems of 2D recognition. Our proposed approach allows, increasing the performance of facial biometrics by offering a system of recognition of facial expressions with 3D image. This system uses a flow composed simultaneously of a color image (2D) and a depth image (3D) captured by the Microsoft Kinect sensor. Our system detects, and identifies the expression of the captured face; then recognize the seven expressions developed by Paul Ekman using the MB-LBP method. We have introduced a new dataset corresponding to 10 human subjects [23, 27] for learning step with a classification using SVM method. The implementation of our solution produces in C⁺⁺ using the OpenCV library.

Keywords: Face recognition, facial expressions, Kinect, MB-LBP, SVM.