This paper introduces, a proposed new identification method based on 3D Ridglet Transform. First phase, it considers the three dimensional fingerprint of human as a Personal Identification Number. Next, it produces the required features using the new proposed 3D Ridglet Transform. This transform is a generalization of adapted 2D Ridglet form. In the second phase we will consider the Back Propagation Neural Network authentication process, the evaluation tests of the proposed algorithm on a given database, for fifteen human Finger-Print, produce a perfect identification results (in comparison with [12]). Based on the evaluation test, we obtain that the authentication of the allowed Human Finger-Print on a noisy data, with a noise level up to 69% also with rotation of the input human Finger-Print up to 9 degree of rotation.