Abstract
In the last years, several researchers have interested in two-dimensional (2D) palmprint recognition. In order to enhance the security of biometric systems, recently, some works proposed to use three-dimensional (3D) palmprint recognition. The advantage of using the 3D capture systems is that they capture the 2D and 3D palmprint at the same time, and they give different and complementary information. The 3D component contains the depth of the palm surface, whereas the 2D component contains the texture. In this paper, we proposed an efficient biometric identification system combining 2D and 3D palmprint by fusing them at matching score level. To exploit the 3D palmprint data, we converted it to grayscale images by using the Mean Curvature (MC) and the Gauss Curvature (GC). Feature extraction is made by a deep learning algorithm that is called the Discrete Cosine Transform Net (DCTNet). The experiments performed on a database containing 8000 samples show that the proposed scheme can achieve a high recognition rate.