Face three-dimensional modeling method based on binocular stereoscopic vision system

Abstract

The invention discloses a human face three-dimensional modeling method based on a binocular stereoscopic vision system. The method comprises the following specific modeling steps: S1, a parallax imaging process; s2, an image preprocessing process; s3, a feature positioning process; s4, a model transformation process; s5, a texture mapping process; s6, a three-dimensional face modeling process; s7, a three-dimensional face module data storage process; according to the invention, a binocular stereo vision system is adopted to extract feature points of a standard three-dimensional face and a target three-dimensional face. Meanwhile, scale normalization and light compensation processing of an image are completed; the alignment operation of all points can be completed at a time, the calculation efficiency is further improved, and the convergence of the resampling process is ensured; feature points are automatically extracted by using an existing feature point extraction method and a grid division method, so that automatic comparison of human faces and data acquisition and storage are realized; the three-dimensional face modeling method is high in automation degree, further improves the fineness of three-dimensional face modeling, and has good market application value.

Description

Technical field [0001] The present invention relates to the field of three-dimensional face modeling, and more particularly to a human face three-dimensional modeling method based on a binocular stereo visual system. Background technique [0002] Three-dimensional (3Dimensions, 3D) face modeling refers to a 3D model of the face of the face from one or more two-dimensional (2Dimensions, 2D) images. [0003] In the relevant technique, the three-dimensional face reconstruction is obtained based on the three-dimensional face model of the target object acquired in the image and the standard facial model in the 3DMMM library (3DMorphable Model, 3D-shaped modified model). By deformation transductionTransfer, DT) technology, use the corresponding relationship between the points and points in the two models (Correspondence) to the point to the high-modest cloud (three-dimensional target object) Face model) The corresponding position is intended to be intended by smooth entry. [0004] Bas...

AI Technical Summary

Problems solved by technology

However, due to the individual differences of faces in the actual process, the 3D face data obtained by acquisition devices such as 3D scanners are not only not aligned in terms of features, but also often have large differences in the number of data points.

Images