Micro-expression recognition method integrating LBP and parallel attention mechanism

Aiming at the problems of weak intensity of facial micro-expression changes, background noise interference and low feature differentiation, a micro-expression recognition network integrating LBP and parallel attention mechanism is proposed. The network inputs RGB images into the densely connected improved Shuffle Stage branch to extract global facial features and enhance the association of contextual semantic information; the LBP image is input into the local texture feature branch composed of multi-scale hierarchical convolutional neural network to extract detail information; after the dual-branch feature extraction, the parallel attention mechanism is introduced at the back end of the network to improve the feature fusion ability, suppress background interference, and focus on the micro-expression feature interest area; the proposed method is tested on three public datasets including CASME, CASME II and SMIC, and the recognition accuracy rates are , and respectively ; the experimental results show that the proposed method effectively improves the accuracy of micro-expression recognition, which is better than many current advanced methods.