Estimation of 3D head region using gait motion for surveillance video

Abstract

Detecting and recognizing people is important in surveillance. Many detection approaches use local information, such as pattern and colour, which can lead to constraints on application such as changes in illumination, low resolution, and camera view point. In this paper we propose a novel method for estimating the 3D head region based on analysing the gait motion derived from the video provided by a single camera. Generally, when a person walks there is known head movement in the vertical direction, regardless of the walking direction. Using this characteristic the gait period is detected using wavelet decomposition and the heel strike position is calculated in 3D space. Then, a 3D gait trajectory model is constructed by non-linear optimization. We evaluate our new approach using the CAVIAR database and show that we can indeed determine the head region to good effect. The contributions of this research include the first use of detecting a face region by using human gait and which has fewer application constraints than many previous approaches.

Fall Detection Based on the Instantaneous Doppler Frequency: A Machine Learning Approach

“…Our assumptions on the speed profile, the person height, the head trajectory during the walk, the maximum speed during the fall, and the duration of the fall are all supported by measurements reported in the literature [14]- [16]. All these facts support the argument that our synthetic data is a reasonable approximation of real-word data.…”

“…The vertical speed v v (t) along the z-axis can be computed by deriving z(t) with respect to t, i.e., v v (t) = dz(t)/dt. It is worth to mention that the validity of the head trajectory model in (3) has been confirmed by fitting it to realworld data obtained from tracking the head trajectory obtained from video recording [16]. We model the static objects in the room, such as the walls and the furniture, by N fixed scatterers S F n (n = 1, 2, .…”

“…The time-variant position z(t) of the scatterer S M (the head) along the z-axis can be written as [16] z(t) = h step cos (2πf step t) + h head (3) where h step , h head , and f step refer to the step height of the head during walking, the person height, and the walking frequency, respectively. The walking frequency can be determined as f step = v h /L s , where L s is the stride length [13].…”

A Machine Learning Approach for Fall Detection Based on the Instantaneous Doppler Frequency

“…1 As the user performs his activity, the position of this scatterer changes over time. In the literature, we find head trajectory models for some activities, such as walking [16]. This trajectory is fed to the RF-activity simulator, which uses the plane wave propagation theory to compute the complex path gain at the receiver.…”

“…where (x M , y M ) denotes the initial position of S M in the x − y plane at time t = 0. The time-variant position z(t) of the scatterer S M (the head) along the z-axis can be written as [16]…”