As the computer vision develops, real-time fall detection based on computer vision has become increasingly popular in recent years. In this paper, a novel real-time indoor fall detection method based on computer vision by using geometric features and convolutional neural network (CNN) is proposed. Gaussian mixture model (GMM) is applied to detect the human target and find out the minimum external elliptical contour. Differently from the traditional fall detection method based on geometric features, we consider the importance of the head in fall detection and propose to use two different ellipses to represent the head and the torso, respectively. Three features including the long and short axis ratio, the orientation angle and the vertical velocity are extracted from the two different ellipses in each frame, respectively, and fused into a motion feature based on time series. In addition, a shallow CNN is applied to find out the correlation between the two elliptic contour features for detecting indoor falls and distinguishing some similar activities. Our novel method can effectively distinguish some similar activities in real time, which cannot be distinguished by some traditional methods based on geometric features, and has a better detection rate.

随着计算机视觉的发展，基于计算机视觉的实时跌倒检测近年来变得越来越流行。提出了一种利用几何特征和卷积神经网络（CNN）的基于计算机视觉的实时室内跌倒检测新方法。高斯混合模型（GMM）用于检测人体目标并找出最小的外部椭圆轮廓。与基于几何特征的传统跌倒检测方法不同，我们考虑了头部在跌倒检测中的重要性，并建议使用两个不同的椭圆分别代表头部和躯干。从每帧中的两个不同的椭圆分别提取长轴和短轴之比，方向角和垂直速度这三个特征，并融合到基于时间序列的运动特征中。此外，使用浅层CNN找出两个椭圆形轮廓特征之间的相关性，以检测室内跌倒并区分一些类似的活动。我们的新方法可以有效地实时区分一些类似的活动，这是传统的基于几何特征的方法无法区分的，并且具有更高的检测率。