Unlock instant, AI-driven research and patent intelligence for your innovation.

Unsupervised cross-library micro-expression recognition method

A recognition method and micro-expression technology, applied in the field of unsupervised cross-database micro-expression recognition, can solve the problems of limited sample recognition effect in the target domain and large differences in the characteristics of the source domain and the target domain, so as to reduce manual labeling work and improve accuracy. rate effect

Active Publication Date: 2021-07-09
SHANDONG UNIV
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In this cross-database situation, the characteristics of the source domain and the target domain are quite different, and the samples of the source domain are directly put into the classifier for training, and the recognition effect on the samples of the target domain is very limited.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Unsupervised cross-library micro-expression recognition method
  • Unsupervised cross-library micro-expression recognition method
  • Unsupervised cross-library micro-expression recognition method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0108] An unsupervised cross-database micro-expression recognition method, such as figure 1 shown, including:

[0109] 1) Prepare training samples, which include macro-expression samples from a macro-expression database and micro-expression samples from a micro-expression database; obtain macro-expression samples, and form a macro-expression data matrix for macro-expression extraction features Among them, d 1 Represents the dimension of macro-expression features, N 1 Represents the number of macro-expression samples; obtain micro-expression samples, and form a micro-expression data matrix for micro-expression extraction features Among them, d 2 Represents the micro-expression feature dimension, N 2 Represents the number of micro-expression samples;

[0110] 2) After data alignment of the macro-expression data matrix and micro-expression data matrix through feature selection and data standardization, the macro-expression data matrix and micro-expression data matrix are r...

Embodiment 2

[0114] According to a kind of unsupervised cross-library micro-expression recognition method described in embodiment 1, its difference is:

[0115] In step 1), the feature extracted for macro-expression is LBP feature. When extracting LBP feature of macro-expression, the face block method of macro-expression adopts the block method of multi-scale LBP-TOP feature, and the feature extracted for micro-expression is multi-scale LBP-TOP features and MDMO features. The block method of multi-scale LBP-TOP features divides the face area into four grid types of 1×1, 2×2, 4×4 and 8×8, with a total of 85 local sub-regions. For each face sub-region Regions are subjected to LBP feature extraction and concatenated.

[0116] In step 2), feature selection is performed on the macro-expression data matrix and micro-expression data matrix respectively by principal component analysis (PCA), and the macro-expression data matrix and micro-expression data matrix are unified into the same dimension ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention relates to an unsupervised cross-library micro-expression recognition method, which comprises the following steps: firstly, aligning and recombining macro-expression data and micro-expression data of a source domain, and selecting an optimal subset which is closer to a target domain as an auxiliary set from a recombined joint data matrix through a source domain selection model; then, matching conditional distributions and edge distributions of the source domain and the target domain dynamically through an adaptive distribution alignment model. And finally, performing reweighting of auxiliary set samples by using L2, 1 norms, reducing the influence of abnormal values, and realizing micro-expression recognition. According to the method, a transfer learning mode is adopted, a macro expression sample having great similarity with micro expressions is used as an assistance, and non-supervised cross-database micro expression recognition is carried out on another micro expression database without any label through a macro expression database with a label and a micro expression database; the time-consuming and labor-consuming manual labeling work of the target domain micro-expression database is reduced, and the micro-expression recognition effect is improved.

Description

technical field [0001] The invention relates to an unsupervised cross-database micro-expression recognition method, which belongs to the technical field of pattern recognition and machine learning. Background technique [0002] As a typical way of non-verbal communication, facial expressions play an important role in human emotion analysis. Microexpressions arise when people try to suppress real facial expressions. Micro-expressions are unconscious, rapid facial expression changes that last from 0.065 seconds to 0.5 seconds. They cannot be freely controlled like macro expressions, and can often reveal the true emotions that people want to hide. Therefore, micro-expression recognition has great application value and development prospects in the fields of criminal investigation and lie detection. [0003] At present, micro-expression recognition generally includes three methods: methods based on artificial features, methods based on deep learning, and methods based on transf...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/46G06K9/62G06N3/04
CPCG06V40/174G06V40/168G06V10/467G06V10/40G06N3/045G06F18/214
Inventor 贲晛烨李冰陈雷肖瑞雪李玉军刘畅
Owner SHANDONG UNIV