Recognition method of facial micro-expression in video sequence

Abstract

The recognition method of human face micro-expression in the video sequence of the present invention relates to the processing that is used to identify the record carrier of figure, is a kind of dynamic space-time texture characteristic method that utilizes HLACLF-TOP algorithm to extract human face micro-expression sequence, and the steps are: human face Micro-expression video Euler amplification; face micro-expression image preprocessing; use HLACLF-TOP algorithm to extract dynamic spatio-temporal texture features of human face micro-expression sequence; use ELM classifier for training and prediction. The method of the invention overcomes the defect in the prior art that it is difficult to recognize micro-expressions due to the small change range of human facial micro-expressions.

Description

technical field [0001] The technical solution of the present invention relates to the processing of a record carrier for recognizing graphics, in particular to the recognition method of human facial micro-expressions in a video sequence. Background technique [0002] Micro-expression (micro-expression) is a very short-lived uncontrollable facial expression revealed by humans when they try to suppress or hide their true emotions. Ekman et al. conducted a series of studies on micro-expression, and the results showed that micro-expression is an effective way to identify lies. Clues can be widely used in security, judicial, clinical and legal fields. However, micro-expressions are short-lived and difficult to recognize. Even with well-trained people, the accuracy rate of micro-expression recognition is only about 40%. Therefore, it is very necessary to develop an automatic micro-expression recognition system and realize computer automatic recognition of micro-expressions, both ...

AI Technical Summary

Problems solved by technology

CN 103258204A discloses an automatic micro-expression recognition method based on Gabor and EOH features, which uses Gabor and EOH features to extract micro-expression features, but the ability of Gabor and EOH to represent the overall situation is weak, and it needs to be combined with the improved GentleSVM classifier to For the classification and recognition of micro-expressions, the performance requirements of the required hardware equipment are high, but the recognition performance is lower than people's expectations

CN104298981A proposes an automatic micro-expression recognition method based on CBP-TOP features, and uses an ELM classifier to classify. This method involves multiplication and power operations when calculating the CBP value of each pixel. Compared with the method of the present invention, the calculation The complexity is high, and the extracted micro-expression information is not complete

Existing face micro-expression recognition methods mainly have the defects of low recognition performance due to the fact that the extracted micro-expression sequence information is single and does not consider multiple frequencies and directions.

Images