Nguyen Hoang Anh, Auvinet Edouard, Mignotte Max, de Guise Jacques A, Meunier Jean
Department of Computer Science and Operations Research, University of Montreal, Montreal, QC H3C 3J7, Canada.
Annu Int Conf IEEE Eng Med Biol Soc. 2012;2012:4835-8. doi: 10.1109/EMBC.2012.6347076.
In this paper, we present a new approach to construct a 3D human skeleton model, which is then used to quantify gait pathologies, using a depth camera. First, thanks to the depth map, we obtain a human depth silhouette in 3D, from which our method is based to estimate each body part position. Second, the angle between the upper and lower legs of the 3D skeleton model is calculated. Finally, we show that using only this angle information is enough to quantify motion asymmetry. This result has been verified through an experimental study with 3 different subjects. Due to its advantages (simple, markerless and low-cost), this method is a promising solution for gait clinics in the future.
在本文中，我们提出了一种构建3D人体骨骼模型的新方法，该模型随后用于使用深度相机量化步态病理学。首先，借助深度图，我们获得3D人体深度轮廓，我们的方法基于此来估计每个身体部位的位置。其次，计算3D骨骼模型中大腿和小腿之间的角度。最后，我们表明仅使用此角度信息就足以量化运动不对称性。这一结果已通过对3名不同受试者的实验研究得到验证。由于其优点（简单、无标记且低成本），该方法是未来步态诊所的一个有前景的解决方案。
