Micro-expression recognition method based on video amplification technology and multi-feature relation model

• Main Authors: WENG XIECHUAN, LIU JUANXIU, YAN BOYUN, LIU LIN, DU XIAOHUI, LIU YONG, ZHANG JING
• Format: Patent
• Language: Chinese; English
• Published: 17.12.2021
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention discloses a micro-expression recognition method based on a video amplification technology and a multi-feature relation model, and belongs to the field of image processing and the field of deep learning. According to the method, firstly, a series of preprocessing operations such as de-noising processing based on a dark channel prior theory, face detection, cutting, face key point detection and face alignment are performed on micro-expression data, the quality of the micro-expression data is improved, then a video amplification network is established, and spatial features of micro-expression frames are extracted, differentiated and amplified. A convolutional neural network structure based on Resnet50 fine tuning is established, and a pre-training model of the convolutional neural network structure on ImageNet is utilized to extract global features of micro-expressions. Then, a feature relation model is built to respectively extract the difference features between amplified micro-expression frames