Micro-expression recognition method based on normal expression assistance

Abstract

The invention discloses a micro-expression recognition method based on normal expression assistance, and the method comprises the steps: 1, carrying out the preprocessing of a micro-expression video and a normal expression video, and constructing a micro-expression data set and a normal expression data set; 2, constructing a micro-expression identity unwrapping network, and extracting micro-expression related features and identity related features from the micro-expression image; 3, constructing a normal expression identity unwrapping network, and extracting normal expression related featuresand identity related features from the normal expression image; and 4, performing joint training on the micro-expression identity unwrapping network and the normal expression identity unwrapping network, and performing fine adjustment on the micro-expression identity unwrapping network by utilizing triple loss, adversarial learning and inequality regularization loss, thereby obtaining an optimal micro-expression identity unwrapping network. According to the method, the deep neural network suitable for micro-expression recognition can be constructed, so that the accuracy and robustness of micro-expression recognition are improved.

Description

technical field [0001] The invention relates to the field of emotion computing, in particular to a micro-expression recognition method based on normal expression assistance. Background technique [0002] As a branch of affective computing, micro-expression recognition has received extensive and sufficient research and attention in recent years. Existing micro-expression recognition methods can be divided into two categories according to the type of features used: methods based on hand-crafted features and methods based on deep features. Histogram of Gradients (HOG), Optical Flow and 3D Orthogonal Planar Local Binary Features (LBP-TOP) are the most commonly used hand-crafted features. With the development of deep learning, more and more fields use deep neural networks to achieve feature extraction. In the field of micro-expressions, there are also many methods based on features extracted by deep neural networks. [0003] However, whether it is a method based on handcrafted ...

AI Technical Summary

Problems solved by technology

[0003] However, whether it is a method based on handcrafted features or a method based on deep features, there are fundamentally unsolvable problems.

For the method based on handcrafted features, because its features can basically only describe shallow features such as texture and timing of micro-expression videos or images, it is not suitable for dealing with complex problems such as micro-expression recognition. Using these shallow features to construct The classifier is less accurate

For methods based on deep neural networks, although the features generated by neural networks are theoretically sufficient for the recognition of micro-expressions, they are based on the premise of sufficient training of neural networks. For complex problems such as micro-expression recognition, neural networks A large amount of training data is needed to be fully trained. However, the existing micro-expression databases only have dozens or hundreds of videos, which are far from meeting the conditions for sufficient training of the corresponding neural network. Therefore, micro-expression recognition based on deep features The method is not very robust

Images