Sleeping Posture Recognition System using Pressure and Infrared Sensors
| Autor: | Tzu-Yu Li, 李子宇 |
|---|---|
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 104 Sleeping posture affects sleeping quality. Physicians can diagnose the patient’s condition based on the sleeping posture. Sleeping posture recognition system can provide sleeping information, which includes sleeping posture, sleeping period, and the number of body turning. Therefore, sleeping posture recognition system can be used to analysis patient’s sleeping quality. Also, it can be used to prevent pressure sores by applying the body turning information. Based on application requirement and cost issue, this proposed sleeping posture recognition system is categorized by two levels: body turning detection level and sleeping quality analysis level. For the body turning detection level, sleeping posture are divided into supine, left lateral and right lateral. The purpose of this recognition level is used for pressure sores prevention in the bed-bound patients. The recognition system of sleeping quality analysis level is more complex because of complicated sleeping posture. However, most of the pressure is distributed on the chest and the hip areas when the human body lying in the bed. It is not cost efficient if sensors are deployed on the whole bed. Therefore, this paper proposes the pressure sensors are deployed on the chest area while developing pressure sensing pad for body turning detection. For the development of sleeping quality analysis level system, the pressure array sensor is deployed to record the pressure distribution of the upper body part, while the lower body area is detected by a single infrared sensor. The proposed scheme is effective and cost efficient for sleeping posture recognition. For the posture recognition algorithm, the fuzzy-based recognition algorithm is proposed in the body turning detection level. While in the sleeping quality analysis level, the fuzzy c-means (FCM) recognition algorithm is proposed to extract the body symmetry property. The experimental result showed that the average posture recognition accuracy for body turning detection can achieve 92.09%. While the accuracy for the sleeping quality analysis can achieve 88.05%. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|