Facial expression recognition method

Abstract

The invention discloses a facial expression recognition method, which belongs to the field of image processing. The facial expression recognition method disclosed by the invention comprises the following steps of: selecting a plurality of neutral expression images of different people and a plurality of various expression pictures of the same people with the neutral expressions from an expression library, and extracting the position information vectors of N characteristic points respectively; obtaining bottom-layer expression characteristic point information vectors feij corresponding to different expression types on the basis of the vector difference of the latter and the former, and training the SVM classifiers of the various expressions on the basis of the feij; conveying the feij each SVM classifier, and calculating the distance from the feij to the hyperplane of each SVM classifier to form an expression characteristic library; extracting the position information vectors of N characteristic points from the neutral expression images of the people to be recognized and the images input in real time and to be recognized, calculating the vector difference of the latter and the former, and conveying the vector differences in each SVM classifier to carry out characteristic conversion processing, so as to obtain a characteristic vector; taking the characteristic vector and a characteristic vector corresponding to an Euclidean distance in the expression characteristic library as the recognized expression. The facial expression recognition method disclosed by the invention is capable of improving the accuracy and real-time performance of recognition.

Description

technical field [0001] The present invention relates to image processing, in particular to facial expression recognition technology. Background technique [0002] Facial expression recognition technology has become a hot development technology in recent years with the rapid development of some related fields such as machine learning, image processing, human recognition, etc. The influence and potential of the facial expression recognition system are extended to a wide range of applications, such as human-computer interaction, intelligent robots, driver status monitoring, and so on. The facial expression recognition system is the premise for computers to understand people's emotions, and it is also an effective way for people to explore and understand intelligence. How to realize the anthropomorphic computer so that it can adaptively provide the most friendly operating environment for the communication object according to the surrounding environment and the state of the obje...

AI Technical Summary

Problems solved by technology

However, the existing problem is that AAM establishes a mathematical statistical model by combining the position information of artificial punctuation points with texture information, which is complex, and then uses optimization theory to locate feature points through multiple iterations, which requires a huge amount of calculation. Accurate but not real-time

Images