Three Dimensional Posed Face Recognition with an Improved Iterative Closest Point Method

Dealing with different head poses is one of the most challenging issues in the area of face recognition. Recently, 3D images have been used for this purpose as they can gather more information from the head area. Kinect was used for capturing 3D images in our research. Iterative Closest Point (ICP) algorithm has been used in many researches to align a rotated pointcloud with its corresponding reference. However it has many variables that can improve its performance. So an improved version of ICP has been introduced in our research and its performance in terms of accuracy and speed has been evaluated. While it can have many applications, we have used it for increasing the performance of posed face recognition. We applied our proposed algorithm on a local database and concluded that it can significantly improve the recognition rate of 3D posed face recognition compared with using original raw posed image. Results of executing the proposed algorithm on a public database also indicate an improvement with respect to other recently proposed methods.