Identification of persons using stereo image pairs and perturbation functions

EasyChair Preprint 1688

Abstract

3D recognition is one of the most progressive methods.  The problem of face recognition is considered. In this model, special points, which form a feature vector, are identified. The method offers the following advantages: continuous and secret identification of the object; it is impossible to use a fake object; twins can be distinguished; weak dependence on head turning (the range of head deflection is substantially increased); weak dependence on external illumination, hair, and face turgidity in the case of a correct choice of the light range. Three-dimensional identification can be used in darkness, and it remains effective even in the case with head turning up to 90◦. For this purpose, a method based on scalar perturbation functions and set-theoretic operation of subtraction is proposed. It is shown that all surface points and the mask volume are used in the process of sample testing for more accurate identification. Using the calibrated stereo pair for the face, the depth map is calculated by the correlation algorithm. As a result, a 3D mask of the face is obtained. Using three antropomorphic masks, a coordinate system that ensures a possibility of superposition of the tested masks is constructed; finally, certain parts are cut off by a clipping plane for equalization of the volumes. Applying the set-theoretic operation of subtraction the set of 3D points (voxels) belonging to the object is determined. This method differs from available 3D methods by the fact that it involves not only all points of the surface in the recognition procedure, but also the volume of the tested mask

Keyphrases: Stereo pair, correlation algorithm, depth map, height map, operation of subtraction, reconstruction, scalar perturbation functions, voxelization

@booklet{EasyChair:1688,
  author    = {Sergey Vyatkin and Alexander Romanyuk and Sergey Pavlov and Oksana Romanyuk},
  title     = {Identification of persons using stereo image pairs and perturbation functions},
  howpublished = {EasyChair Preprint 1688},
  year      = {EasyChair, 2019}}