Posture Detection System Based On Dense Pressure Sensor

During the average seven to nine hours every day that we sleep,the human body heals itself and grows.Those who have poor sleeping quality are prone to stress,fatigue,attention deficit,or eating disorders.Poor sleeping posture has bad influence on quality of sleep.One of the most common conditions is sleep apnea.In addition to direct effects on respiratory problems,sleeping posture can be important for patients to recovery from serious operative procedures.It has been shown that sleep quality may affect the recovery time of patients in hospital.More specifically,one of the main problems for post-surgical patients and elderly patients is formation of pressure ulcers.Therefore,sleeping position recognition has very important research and application value.In this dissertation,we mainly use a posture detection system and image recognition algorithm,which based on dense pressure sensor arrays to detect body's postures including sitting postures and sleeping postures.In this dissertation,we mainly detect and recognize the human body's postures in sleep state.According to the principle of the resistive pressure sensor,we designed the dense pressure sensor arrays with a relatively high resolution.And we also did some performances testing on kinds of sensitive materials to select right sensitive material.On the static images of sleeping position recognition,we focused on recognizing two sleeping postures,supine and fetus.Preliminarily,we preprocessed images to improve sleeping image quality.And then,we recognized the sleeping postures using Phash algorithm.The posture detection system mainly consists of two parts,hardware and software.we proposed a framework for sleeping posture detection based on a dense pressure sensitive bedsheet prototype(48×20 sensors)designed using electronic textile material.To make full use of resources,we designed two levels of multiple analog switch selection circuits to select the signals sensed by the sensing units at high speed.A/D conversion module of the signal acquisition uses ALIENTEK Mini STM32 in the A/D conversion channel.The system uses serial RS232 communication module to make communication with the computer.There are two main parts of the software.The part of A/D sampling programs is used to complete the data acquisition.For the part of posture detection algorithm,we firstly employed Canny edge detector to remove noise and extract useful data,and then,we used perceptual hashing algorithm to recognize the posture images.