Facial expression recognition method based on ELM

Abstract

The invention discloses a facial expression recognition method based on an ELM. The method comprises the steps that a facial expression image is processed to obtain a facial region, image segmentation is conducted on an expression region containing expression features to obtain an eye expression region body, a nose expression region body and a lip expression region body, then Gabor filters are used for conducting expression feature extraction on the expression region bodies to obtain a total feature vector of each expression, the obtained expression feature vectors are used for training ELM models, the ELM models for all the expressions are combined to form an ELM expression classifier, and at last, the ELM classifier is tested to achieve the purpose of expression classified recognition. The facial expression recognition method based on the ELM increases the expression recognition rate on the condition that illumination influences are avoided.

Description

technical field [0001] The invention relates to the field of human facial expression recognition, in particular to an ELM-based human facial expression recognition method. Background technique [0002] Facial expression recognition is an important research direction in human-computer interaction research, involving image processing and analysis, artificial intelligence, pattern recognition, computer vision, computer graphics and other disciplines. The study of human-computer interaction is also a promotion of these disciplines and related fields. In the multi-mode human-computer interaction interface, the combination of facial expressions, sight, posture, and voice can achieve more efficient human-computer communication. [0003] In-depth research on expression recognition can enable robots to better understand human emotions and psychology, and become more humane and intelligent in the interactive environment with people, which is more conducive to serving humans, and faci...

AI Technical Summary

Problems solved by technology

The feature point tracking method and the model tracking method respectively obtain the expression feature parameters of the sequence by tracking predetermined feature points or models, both of which require manual intervention and are difficult to implement automatically

The optical flow feature method analyzes expressions based on the sequence of optical flow changes. The disadvantage is that it is very sensitive to illumination changes.

[0008] The existing expression classification methods mainly include hidden Markov model (HMM), neural network, Adaboost, support vector machine (SVM), but these methods also have slow training speed and low generalization performance in expression training classification. The problem of many adjustment parameters

Images