利用主成分分析法之三維傾角計算應用於多人跌倒偵測

Fall detection for multiple persons using a PCA approach to 3D inclination

吳建峰(Chien-Feng Wu)

指導教授 ： 林錫寬

國立交通大學/電機學院/電機與控制工程系所/博士(2020年)

摘要

當前世界各已開發國家都已經漸漸邁入老人化社會，『老人照護』議題，特別是跌倒偵測方法也被普遍的研究，本研究主要的跌倒偵測方法分為兩個部份，第一個部份為3D物件的重構：利用座標轉換的方法將投影座標轉換成為大地座標，其中經由本研究推導之主成份分析法 (Principle component analysis, PCA)，配合雷射水平儀（Laser Level），對轉換座標的參數進行校正。第二個部份利用主成份分析法進行正交回歸計算得到3D物件的傾角。最後本論文將上述主要方法實現於多人跌倒偵測系統上，進行此方法在跌倒偵測系統上的可行性評估，本方法準確度(95.94％)和敏感度(94.44%)，其表現皆優於相關文獻。

關鍵字

跌倒偵測 ； 主成分分析法 ； 三維傾角

並列摘要

At present, all developed countries in the world have gradually entered an aging society. The issue of "elderly care", especially the fall detection method, has also been widely studied. The main fall detection method in this study is divided into two parts. The first part is the 3D objects reconstruction: the transformation of projection coordinates into geodetic coordinates using the method of coordinate conversion, in which the principal component analysis method derived from this study (Principle component analysis, PCA), with laser level, to correct the parameters of the conversion coordinates. The second part uses PCA to perform the orthogonal regression calculations to obtain the 3D object inclination. Finally, this paper implements a multi-person fall detection system. The feasibility evaluation of this method on the fall detection system is carried out. The accuracy (95.94%) and sensitivity (94.44%) of this method are both better than related literature.

並列關鍵字

Fall detection ； Principal Component Analysis ； 3D inclination

參考文獻

[1] United Nations, “World Population Prospects 2019”, United Nations, 2019.

[2] World Health Organization, “WHO global report on falls prevention in older age,” World Health Organization, 2018.

[3] S.W. Yang, and S.K. Lin, “Fall detection for multiple persons using depth image processing technique,” Computer Methods and Programs in Biomedicine, vol.114, no.2, pp. 172-182, 2014.

[4] D. C. Lay, Linear Algebra and its Applications, 5th Edition, Addison Wesley, 2016.

[5] M. Mubashir, L. Shao and L. Seed, “A survey on fall detection: Principles and approaches,” Neurocomputing 100, pp. 144-152, 2013.