Researches On The Sleep Position Recognition Method And Related Technologies By Using A Capacitive Flexible Pressure Array

The research on the recognition of human sleeping position is one of the key directions in the field of physiological information monitoring and pattern recognition research in recent years.Incorrect sleep posture can affect the psychological and physical health of modern people,especially those with sleep disorders.Therefore,a good sleep posture is one of the key indicators for assessing modern people's sleep quality and prevention and treatment of related diseases.At present,the recognition of sleep posture is mostly based on computer vision.The method used is restrained by many people and is expensive,which is not conducive to mass popularization.At present,relevant research fields at home and abroad are seeking a method of unconstrained monitoring and recognizing sleeping position,most of which are in the primary research stage.This project aims to develop and design a capacitor array flexible pressure sensor as the key technology,and deeply research the sleeping position recognition algorithm based on this sensor.It mainly includes the design of the main part of the flexible pressure sensor and the multi-class sleep based on SVM(Support Vector Machine).The development and testing of the pose classification algorithm,the specific work content is as follows:1.The main part of the flexible pressure sensor design: designed to design a capacitive array flexible pressure sensor that can be applied to the information collection and monitoring of human sleeping position information,including the material selection and structural design of the sensing element,and the measurement component to the capacitive signal.Hardware and software technology and methods for acquisition,transformation and processing.2.Test and analysis of the manufactured sensor: The flexible pressure sensor produced by the actual test can complete the task of collecting the sleeping position pressure information of the human body,including the main parameter test of the sensor performance,the experimental test under the three postures of the human hand,the foot and the back lying.3.Generation and preprocessing of sleeping position image: The set of sleeping position picture samples designed to be the specification of sleeping position recognition algorithm proposed in this paper,including the generation of sleeping position gray level map,threshold segmentation,and pretreatment of sleeping position image.4.The development and testing of multi-class sleeping position classification algorithm based on SVM: This paper proposes a multi-class(six)sleeping position classification algorithm based on SVM multi-class classification problem,including solving multi-class classification in training and recognition stage.The adopted scheme,as well as the optimization problem of SVM kernel function parameters and penalty factor C,are compared with other sleeping position recognition algorithms in terms of recognition accuracy and number of sleeping postures.