Gait Analysis and Human Motion Tracking - Intelligent Video Event Analysis and Understanding - page 39

Gait Analysis and Human Motion Tracking

Huiyu Zhou

Abstract. We present a strategy based on human gait to achieve efficient tracking,
recovery of ego-motion and 3-D reconstruction from an image sequence acquired
by a single camera attached to a pedestrian. In the first phase, the parameters of the
human gait are established by a classical frame-by-frame analysis, using an gener-
alised least squares (GLS) technique. The gait model is non-linear, represented by a
truncated Fourier series. In the second phase, this gait model is employed within a
“predict-correct" framework using a maximum a posteriori, expectation maximiza-
tion (MAP-EM) strategy to obtain robust estimates of the ego-motion and scene
structure, while continuously refining the gait model. Experiments on synthetic and
real image sequences show that the use of the gait model results in more efficient
tracking. This is demonstrated by improved matching and retention of features, and
a reduction in execution time, when processing video sequences.

1 Introduction

One of the applications of closed-circuit television (CCTV) cameras is in surveil-
lance systems that provide monitoring information in order to protect civilians
against theft, vandalism and crime [31, 32, 37, 12]. Most of these cameras, to our
knowledge, are statically installed in commercial and residential sectors such as
schools, banks and parking places. Unfortunately, these cameras can only cover a
part of the entire scene, and cannot track subjects that exit the field of view of the
cameras. In contrast, portable camera systems can be used to monitor a scene that
changes in the field of view of the officer who carries the camera. To conduct effec-
tive video analysis, one of the main challenges is efficient recovery of 3-D structure
of the scenes and the motion trajectory of the camera, based on 2-D image inputs.
Estimating 3-D structure and camera motion from 2-D images is one of the
fundamental problems in computer vision, e.g., [29, 26]. This problem has been