Biometric Feature Extraction And Identification Based On People-oriented Intelligent Sense

With the development of sensor technology and computer technology,artificial intelligence has been applied to all walks of life.However,in some complex scenarios,artificial intelligence still has problems such as excessive dependence on algorithms,inflexible human-computer interaction,and inability to adapt to different people and different states of people.Therefore,to be people-oriented has become one of the main directions of artificial intelligence development.The machine actively perceives people's physiological state,emotion and identity is a new dimension of artificial intelligence research,which adapts to different people and adapts to different states of people.Focusing on the machine's active sense of human physiological state,emotion and identity,the paper proposes the concept of "people-oriented Intelligent Sense".Based on people-oriented Intelligent Sense,the algorithm not only can sense human body state information by the abilities to analyze and calculate,but also considers higher-dimensional information about people such as emotions and identities in decision-making.Based on the concept of people-oriented Intelligent Sense,this paper focuses on the extraction of biological features to sense the human state,and studies the machine's active perception of human physiological state,emotion and identity.In response to emotional perception,the paper uses a single video or audio sensor to replace complex dedicated sensors.Video-based non-contact heart rate acquisition,video-based blood pressure acquisition,and audio-based neonatal crying inherent emotional feature extraction and analysis are realized.A video-based inquiry system is designed.A research method that is based on small sample sets,non-learning,and can extract inherent explainable stable features is proposed.In response to identity perception,the paper uses multiple cameras and the individual specific principle of pulse to realize recognizing and positioning multiple people with multiple cameras based on "vibration signals" of the skin surface.A principle of the difference and synchronism of the "vibration signals" under multiple cameras is proposed.