Three dimensional human face recognition method based on intermediate frequency information in geometry image

摘要

The present invention relates to a three-dimensional face identification method based on the information of the medium frequency geometrical image, the procedure shown below. (1) intended to pre-processing for the library collection models and test models of three-dimensional face, a three-dimensional face, including the area division, smoothed and point cloud dilution end, remove the set of points around the mouth and hold the upper half of the face do. (2) through a mesh parameterization mapping a two-dimensional mesh in the upper half of the face, and the linear interpolation the three-dimensional coordinates of the mesh vertices to get the three-dimensional coordinates of each pixel attribute point generates a geometric image of the three-dimensional face model. (3) a multi-scale multi-scale Haar filters for the geometric image using a wavelet filter, and extract the medium frequency horizontal image information, the image information and a vertical medium wave medium frequency diagonal information in the facial expression image samneunda immutable feature of the face. (4) using a wavelet domain structure similarity algorithm to calculate the similarity of the test model and the model library collection. (5) The model library collection having a maximum degree of similarity is determined to belong to the same object as the test model according to the test model and the degree of similarity for each of the three-dimensional face model library collection library collection. ;