Micro-expression recognition method based on video motion amplification and optical flow features

• Main Authors: ZHAO MINGHUA, DU SHUANGLI, LI PENG, DONG SHUANGSHUANG, WANG LI, WANG LIN, HU JING
• Format: Patent
• Language: Chinese; English
• Published: 11.11.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention discloses a micro-expression recognition method based on video motion amplification and optical flow features. The method specifically comprises the steps that a data set is selected and classified according to emotion; all original image frame sequences of the selected data set are preprocessed, and all obtained single-channel grey-scale map sequences serve as one part of network model input; an RAFT network structure based on deep learning is adopted to calculate optical flow features of all image frame sequences, and an optical flow graph obtained through visualization is used as the other part of network model input; and superposing all the single-channel grey-scale map sequences and all the visual RGB optical flow map sequences to form a four-channel image, inputting the four-channel image into a designed VGG16 network to extract spatial domain features of micro-expressions, and classifying the spatial domain features to obtain final recognition precision. According to the method, the key p