以深度攝影機應用於遠距復健評估之三維人體影像骨架化精確度改善之研究

Improving Tracking Accuracy on Skeletonization for 3D Human Image Applied in Telerehabilitation Assessment using Depth Camera

許炳銓(Bing-Chuan Shiu)

指導教授 ： 張淵仁

中臺科技大學/健康科學院/醫學工程暨材料研究所/碩士(2012年)

摘要

遠端評估病患復健動作使物理治療師了解病患復健成效是遠距復健重要架構之一，本研究使用深度攝影機Kinect，參考過去學者透過影像骨架化的方式分析人體動作，綜合過去學者方法，利用一個簡單的校準姿勢，透過輪廓判斷找出關節點將其連線形成骨架，因為人體任何運動都由關節做支撐，因此可透過關節點了解人體的動作，並利用Kinect即時的產生三維座標資料，使物理治療師能夠即時的了解病患的動作。
本研究實際以物理治療師在評估病患復健動作之中風評估量表(STREAM)驗證本研究所提出復健評估系統之準確性，並根據物理治療師建議的常用六項動作，比較本研究之骨架關節點與商業動作捕捉系統Vicon Nexus之信效度測試，並利用運動學(Kinematics)之軌跡進行Kinect動作修正因光線問題而產生Kinect偵測失準現象，以提升系統偵測人體動作之準確定，進一步改善物理治療師執行遠距復健對病患動作評估準確性。本研究使用深度攝影機建立一套遠距復健評估系統，證明可以有效改善遠距復健評估之動作真確準確性，對未來遠距復健系統之實用性有極大助益。

關鍵字

動作評估 ； 深度攝影機 ； Kinect ； 骨架化 ； 遠距復健 ； 中風評估量表STREAM

並列摘要

This paper presents a methodology to assist the Tele-rehabilitation process of a stroke patient by evaluating human motions, in this study, rehabilitation assessment used the Microsoft depth camera Kinect, which can record three-dimensional coordinates of human motion. Kinect can compare a patient’s motion with a gold standard and calculate its accuracy. A simple position correction algorithm was used to improve the accuracy of tracking human movement acquired from Kinect. This system reliability and validity levels of significance less than 0.01, This method provides a possibility for stroke patients to perform Tele-rehabilitation through the Internet using Kinect, as suggested in this study. Stroke patients can benefit by saving time and manpower for travel to and from the hospital. Moreover, the rehabilitation process becomes more interesting by using game-based rehabilitation programs with Kinect.
We use stroke rehabilitation assessment of movement (STREAM) be experimental action, compare our system the reliability and validity testing of skeleton joint points with the commercial motion capture system Vicon Nexus, and we use kinematics improve Kinect detect distortion by light interference, improve the therapist to perform the assessment of the accuracy of the Tele-rehabilitation patients actions, practicality of the telephoto rehabilitation system has proved to be very useful, such treatment can reduce the cost of rehabilitation, which includes travel time between home and hospital, manpower for rehabilitation evaluation, and registration fee for outpatient services.

並列關鍵字

Microsoft Kinect ； Tele-rehabilitation ； Stroke ； Human motion ； Argument reality ； Evaluation of human motion ； VICON ； Technology acceptance model

參考文獻

[26] 黃曉薇, “應用在遠距復健的人體運動評估之初步研究,” 國立中興大學資訊科學與工程學系碩士學位論文.

[1] Aphrodite, G., Fabrice, C., Howard, T., “Real-time 3-D human body tracking using learnt models of behavior,” Computer Vision and Image Understanding 112–125 2008.

[2] Atif, A., Jongeun, C., Abdulmotaleb, E., S., “AR-REHAB: An Augmented Reality Framework for Poststroke-Patient Rehabilitation,” IEEE transactions on instrumentation and measurement, IEEE, 2010.

[3] Alexandros, A., C., Pau C., P., Francisco F., R., “A review on vision techniques applied to Human Behaviour Analysis for Ambient-Assisted Living,” Expert Systems with Applications 39, 10873–10888, 2012.

[6] Diarmaid, F., Dan, K., Tomas, W., Charles, M., Brian C., “Usability Evaluation of E-Motion: A Virtual Rehabilitation System Designed to Demonstrate, Instruct and Monitor a Therapeutic Exercise Programme,” Proceedings of Virtual Rehabilitation, 2008.