Dynamic Distance-Based Shape Features for Gait Recognition

Abstract

We propose a novel skeleton-based approach to gait recognition using our Skeleton Variance Image. The core of our approach consists of employing the screened Poisson equation to construct a family of smooth distance functions associated with a given shape. The screened Poisson distance function approximation nicely absorbs and is relatively stable to shape boundary perturbations which allows us to define a rough shape skeleton. We demonstrate how our Skeleton Variance Image is a powerful gait cycle descriptor leading to a significant improvement over the existing state of the art gait recognition rate.

Lidar-Based Gait Analysis and Activity Recognition in a 4D Surveillance System

“…Figures a)-e) demonstrate the steps of Lidar based moving object detection and multi-target tracking, f)-h) shows the reconstruction process of moving avatars in the 4D studio, i) displays a reference video image from the same scenario (not used by the i4D workflow) and j) is a snapshot from the reconstructed 4D scenario, shown from four different viewpoints A possible option for obtaining depth information from the scene is using stereo cameras or Time-of-Flight (ToF) technologies. Cheap Kinect sensors have been investigated for gait analysis in a number of works [30], [31], [32], and a corresponding gait database has already been published [33] for reference. However Kinects are still less efficient for applications for real life outdoor scenarios due to their small FoV and range (resolvable depth is between 0.8m -4.0m), and the low quality outdoor performance of the sensor, especially in direct sunlight.…”

Gait Recognition with Compact Lidar Sensors

“…There are many gait recognition approaches published in the literature which are based on point clouds (Tang et al, 2014;Gabel et al, 2012;Whytock et al, 2014;Hofmann et al, 2012), yet they use the widely adopted Kinect sensor which has limited range and a small field-of-view and is less efficient for applications in real life outdoor scenarios than LiDAR sensors. Also the Kinect provides magnitudes higher density than an RMB LiDAR, so the effectiveness of these approaches are questionable in our case.…”

On covariate factor detection and removal for robust gait recognition

“…Manifold single compact 2D gait representations exist promoting static features (torso), dynamic features (limb motion) [14,17,19,33,35] or a combination thereof [1,2,33,38]; those containing only dynamic features tend to be naturally robust given their saliency [25] over time compared to only static features. Regardless of representations, misclassification occurs from neglecting the following: (a) covariate factor pixel-wise confusion with natural gait motion and (b) the degree, severity and uniqueness in which covariate factors affect gait appearance and motion.…”

“…Our bolt-on module is applied to the Gait Energy Image [14], Gait Variance Image [33], Skeleton Energy Image [33] and Skeleton Variance Image [33] which vary in feature content and natural robustness, demonstrated in Fig. 1.…”