Time–frequency fusion learning for photoplethysmography biometric recognition

Photoplethysmography (PPG) signal is a novel biometric trait related to the identity of people; many time- and frequency-domain methods for PPG recognition have been proposed. However, existing domain only consider single or feature-level fusion time frequency domains, without considering exploration correlations domains. The authors propose time–frequency method with collective matrix factorisation (TFCMF) that leverages learn shared latent semantic space by exploring In addition, utilise ℓ2,1 norm constrain reconstruction error matrix, which can alleviate influence noise intra-class variation, ensure robustness learnt space. Experiments demonstrate TFCMF has better performance than current state-of-the-art recognition.