基於長時間穿戴感測器之健身房身分識別

Person Identification in Gym Based on Data of Long-Term Wearable Sensors

黃睿騏(Jui-Chi Huang)

指導教授 ： 荊宇泰

國立交通大學/資訊學院/資訊科學與工程研究所/碩士(2018年)

摘要

本篇論文利用穿戴於右手臂、右大腿及右腳踝感測器所收集的三軸加速度和角速度資料，目標是建置出一個包含兩種模型的系統，最後達到能在一段健身房內的資料中識別出使用者是誰，同時也知道使用者在做什麼運動的功能。
研究的方法分成三個主要階段，首先第一階段藉由收集各種不同健身房動作資料加以處理並擷取特徵，建立出動作辨識模型。第二階段則是使用不同人走路的資料，切割出步態循環後使用步態對時間上的變化當作特徵建立出身份辨識模型。而為了解決不能分辨不屬於模型中的人，提出了另外一種分法，僅使用屬於本人的資料，透過將特徵轉換成距離的方式建立身份識別模型。
最後透過代入實際上的健身房長時間資料，證實這樣的系統在長時間的健身房資料中可以準確達到前面所說的功能。

關鍵字

身份辨識 ； 動作辨識 ； 穿戴式裝置 ； 健身房運動 ； 長時間資料

並列摘要

This article is based on analyzing the gyroscope and acceleration data from three wearable sensors at right arm, right thigh and right ankle. Our goal is to build a system containing two kinds of models that can know what the user's doing in the gym and also who he or she is.
This research is mainly separated by three parts. In the beginning, we construct an activity recognition model using time and frequency domain features from data of gym activities. In the second part, we first split different people's walking data to get gait cycle. Then use the gait cycle to extract features represent the variation when people walking and build the identification model. Also, aiming to solve the problem that we cannot identify people not in the model, we change the features to distance based.
In the final part, we testify the system using long-term Gym data including different activities. The result shows that this system reach our goal with high accuracy.

並列關鍵字

Person Identification ； activity recognition ； wearable device ； Gym Activities ； Long-term data

參考文獻

[1] Y. Lu, A. Fleury, J. Boonaert, S. Lecoeuche, and S. Ambellouis. Online person identification and new person discovery using appearance features. In 2015 IEEE International ConferenceonEvolvingandAdaptiveIntelligentSystems(EAIS),pages1–8,Dec2015.

[2] N.Radha,A.Shahina,andA.NayeemullaKhan. Apersonidentificationsystemcombining recognitionoffaceandlip-readpasswords.In2015InternationalConferenceonComputing andNetworkCommunications(CoCoNet),pages882–885,Dec2015.

[3] Shugang Zhang, Zhiqiang Wei, Jie Nie, Lei Huang, Shuang Wang, and Zhen Li. A review on human activity recognition using vision-based method. In Journal of healthcare engineering,2017.

[4] ThingomBishalSingha,RajsekharKumarNath,andAVNarsimhadhan.Personrecognition usingsmartphones’accelerometerdata. arXivpreprintarXiv:1711.04689,2017.

[5] Hoang Minh Thang, Vo Quang Viet, Nguyen Dinh Thuc, and Deokjai Choi. Gait identification using accelerometer on mobile phone. In Control, Automation and Information Sciences(ICCAIS),2012InternationalConferenceon,pages344–348.IEEE,2012.