Micro-facial expression detection method

Abstract

The present invention provides a micro-facial expression detection method. The method comprises: extracting N feature points from each image frame in a to-be-detected image sequence and aligning human faces; selecting K key feature points from the N feature points and dividing the K key feature points into M feature point clusters; selecting to-be-detected clusters from the M feature point clusters; selecting a base frame; calculating key point vectors of the to-be-detected clusters in each image frame; taking the key point vector of the base frame as a base vector, and calculating a deformation vector of each image frame; taking D times a maximum deformation vector as a deformation threshold; adding all image frames with the deformation vectors greater than the deformation threshold into the quasi-micro-facial expression frame sequence; and when there are consecutive frames in the quasi-micro-facial expression frame sequence, and the number of frames is greater than or equal to the preset frame threshold, taking the continuous frames as the micro-expression frame sequence. According to the method provided by the present invention, the micro-facial expression required to be detected is detected from the to-be-detected image sequence, and the recognition capability of the micro-facial expression and the robustness of the detection method are greatly improved.

Description

technical field [0001] The present application relates to the technical fields of pattern recognition and computer vision, and in particular to a method for detecting micro-expressions. Background technique [0002] In recent years, the technology of realizing human-computer communication through emotion recognition of voice, facial expression, body language and other features has developed rapidly. Among them, facial expressions play an important role in analyzing human emotions, however, in many situations people hide or suppress their true emotions. [0003] Micro-expression is a very fast expression that lasts only 1 / 25 second to 1 / 5 second, it can reveal the true emotion that human beings try to hide, so in national security, clinical diagnosis, case detection, danger warning, personal defense And other fields have shown good application prospects, especially in lie detection has important application value. However, the research on micro-expressions started relativel...

AI Technical Summary

Problems solved by technology

[0007] However, there are still some problems in the micro-expression detection method in the prior art, the accuracy of the detection result is still not very high, and the automatic recognition ability is still not very strong

Images