The Estimation of Heights and Occupied Areas of Humans from Two Orthogonal Views for Fall Detection

Abstract

In this paper, we present a video-based method of detecting fall incidents of the elderly living alone. We propose using the measures of humans' heights and occupied areas to distinguish three typical states of humans: standing, sitting, and lying. Two relatively orthogonal views are utilized, in turn, simplifying the estimation of occupied areas as the product of widths of the same person, observed in two cameras. However, the feature estimation based on sizes of silhouettes varies across the viewing window due to the camera perspective. To deal with it, we suggest using Local Empirical Templates (LET) that are defined as the sizes of standing people in local image patches. Two important characteristics of LET are: (1) LET in unknown scenes can be easily extracted by an automatic manner, and (2) by its nature, LET hold the perspective information that can be used for feature normalization. The normalization process is not only to cancel the perspective but also to take the features of standing people as the baselines. We realize that heights of standing people are greater than that of sitting and lying people. People in standing states also occupy smaller areas than whom in sitting and lying states. Thus, three humans' states fall into three separable regions of the proposed feature space, composing of normalized heights and normalized occupied areas. Fall incidents can be inferred from time-series analysis of human state transition. We test the performance of our method on 24 video samples in Multi-view Fall Dataset (1) leading to high detection rates and low false alarms, which outperform the state-of-the-art methods (2) (3) tested on the same benchmark dataset.

Detecting Fall Incidents of the Elderly Based on Human-Ground Contact Areas

“…cameras 2 and 7 (positioned 180 o apart), cameras 2 and Table I. Although results in [6] seem to be better than ours, it does not consider a fall from sitting to lying states that frequently happens with the elderly. Our performance is competitive with low computational cost.…”

“…We realize that [5], [6], [18] based on 3D human reconstruction have good performance. However, it seems to be not necessary to reconstruct 3D human models.…”

“…3D silhouettes [5] are reconstructed from multiviews since its vertical distributions are different for standing and lying people. Hung and Saito [6] approximate people by a rectangular cuboid by two cameras whose fields of view are relatively orthogonal. A feature space composing of height and bottom area of cuboid is studied to classify human states into standing, sitting and lying for event detection.…”

“…We adopt the common metric used in state-of-the-art methods [5], [6], [7] for a fair performance comparison by computing sensitivity and specificity,…”

“…It leads to the fair comparison with state-of-the-art methods [5], [6], [7], tested on the same dataset. Our results are competitive.…”