Three dimensional facial expression recognition method based on multiple dimensional characteristics of representative regions

Abstract

The invention provides a three dimensional facial expression recognition method based on multiple dimensional characteristics of representative regions. The three dimensional facial expression recognition method includes the following steps: performing pretreatment operation on the three dimensional facial data, and acquiring the more normalized point cloud data; performing automatic calibration of a representative region of three dimensional facial expression; according to the position of the tip point of nose, completing automatic calibration of the eye region (E region), the nose region (N region) and the mouth region (M region); respectively extracting the three dimensional characteristics and the two dimensional characteristics of three representative regions, and performing gauss normalization and fusion of characteristics; and at last, according to the fusion characteristics of each representative region, performing SVM training to realize three dimensional facial expression recognition. The three dimensional facial expression recognition method based on multiple dimensional characteristics of representative regions not only can observe the contribution degree of different regions of the face on different expressions, but also can effectively recognize different expressions of the three dimensional face.

Description

technical field [0001] The invention relates to the fields of computer vision and pattern recognition, in particular to a three-dimensional facial expression recognition method based on multi-dimensional features of typical regions. Background technique [0002] Facial expression recognition is a hot research issue in the field of computer vision and pattern recognition, and has been paid more and more attention by researchers in the 30 years since the 1980s. At the same time, as an important branch of artificial intelligence and emotion analysis, facial expression recognition has great research value and application prospects in the fields of human-computer interaction and intensive care unit monitoring. [0003] Early facial expression recognition focused on two-dimensional images and image sequences, that is, the implementation of algorithms on videos. However, since a 2D face image is a 2D plane projection of a 3D object, features such as texture and shape will inevitab...

AI Technical Summary

Problems solved by technology

However, since the 2D face image is a 2D plane projection of a 3D object, it will inevitably lose texture, shape and other features during the projection process, and will also be affected by posture, lighting, etc.

Images