Unrestricted pattern selection in contactless palmprint identification systems

The biggest problem in the contactless biometry recognition system, whose popularity is increasing recently, is image distortion resulting from pose freedom (different angles and distances). In this study, a new approach is presented on the palm stance order to increase success on contactless palmprint verification systems. 3D palm is obtained through the stereo images received from the 2 cameras placed at different locations. In the process of camera calibration, internal and external parameters of the camera are obtained and with the help of these parameters different images that are taken from two cameras are projected on the same plane with the stereo rectification process. With stereo matching process, the key-points in rectified images individually identified with SIFT and SURF methods. With three-dimensional reconstruction stage positions of these points in 3-dimensional space are obtained. Then 3-dimensional plane which is formed by these points and the direction of this plane is determined. Finally, after adjusting the palm's pose, palmprint recognition process is carried out with the help of the Active Appearance Model. The results obtained in this study are given as images that demonstrates the robustness of the method.