3D FACE RECOGNITION METHOD BASED ON INTERMEDIATE FREQUENCY INFORMATION IN GEOMETRIC IMAGE

摘要

A 3D face recognition method based on intermediate frequency information in a geometric image. The steps are as follows: (1) preprocessing a library set model of a 3D face and a test model, including 3D face area cutting, smoothing processing and point cloud thinning, and at last cutting the point sets near the mouth and retaining the upper half of the face; (2) mapping the upper half of the face to a 2D grid through grid parameters, and performing linear interpolation on the 3D coordinates of the grid top to acquire the 3D coordinate attributes of each pixel point and generate a geometric image of a 3D face model; (3) performing multi-scale filtering on the geometric image with a multi-scale Haar wavelet filter to extract a horizontal intermediate frequency information image, a vertical intermediate frequency information image and an diagonal intermediate frequency information image as invariable facial expression features of the face; (4) calculating the similarity between the test model and the library set model with a wavelet domain structuring similarity algorithm; and (5) judging that the library set model and the test model with the maximum similarity belong to the same body according to the similarity between the test model and each library set model of the 3D face library set.