Gait Recognition Using Image Self-Similarity

Abstract

Gait is one of the few biometrics that can be measured at a distance, and is hence useful for passive surveillance as well as biometric applications. Gait recognition research is still at its infancy, however, and we have yet to solve the fundamental issue of finding gait features which at once have sufficient discrimination power and can be extracted robustly and accurately from low-resolution video. This paper describes a novel gait recognition technique based on the image self-similarity of a walking person. We contend that the similarity plot encodes a projection of gait dynamics. It is also correspondence-free, robust to segmentation noise, and works well with low-resolution video. The method is tested on multiple data sets of varying sizes and degrees of difficulty. Performance is best for fronto-parallel viewpoints, whereby a recognition rate of 98% is achieved for a data set of 6 people, and 70% for a data set of 54 people

Gait recognition without subject cooperation

“…BenAbdelkader et. al [3] propose to represent gait using image self similarity which measures the similarity between pairs of silhouettes in a gait sequence. It is claimed that the self similarity representation of gait encodes a projection of gait dynamics and is resistant to noise.…”

Gait recognition based on shape and motion analysis of silhouette contours

“…The methods in [4,10,11,16,20,21,22,41,42] achieve limited invariance to a few covariate factors for improved identification rate. The method in [10] uses appearance and dynamic traits of gait by analysing parameters of ellipses fitted to seven regions of a subject's silhouette, i.e., centroid, aspect ratio and elongation along with the subject's height for identification which is invariant to limited clothing variations and segmentation imperfections.…”

“…The performance of STS-DM on the lateral view of silhouettes of the CMU MoBo data set is compared with shape sequence matching (SSM) based method in [17], SSP [16], STM-SPP [23] and SVB frieze [33]. Table 2 shows that the shape based approach in SSM using stance correlation for the subjects walking parallel to the image plane is robust to variations in walking speed, but its performance degrades significantly when the shape of the silhouettes change due to different activities (e.g., fast walk vs walking with ball).…”

“…However, they are characterised by complex search and mapping processes. The model-free methods (e.g., [4,13,14,15,16,17,18,19,20,21,22,23]) analyse the spatio-temporal shape and motion characteristics of a subject's silhouettes without assuming any explicit model of the subject's body. Although the inter-subject discriminability of these methods are high, they are susceptible to variations in viewpoints and the subject's attire.…”

Hybrid Dynamical Models of Human Motion for the Recognition of Human Gaits

“…In all cases the first step consists in deriving a compact representation of the motion, such as binary silhouettes (Sarkar et al 2005;Kale et al 2004), optical flow (Little and Boyd 1998), joint angles of an articulated body model with image-based tracking (Bregler 1997;Bissacco et al 2001;North et al 2000), or other spatio-temporal motion descriptors (BenAbdelkader et al 2004;Efros et al 2003;Zelnik-Manor and Irani 2006). Then some statistics are computed on the reduced data and pattern recognition techniques such as principal component analysis (BenAbdelkader et al 2004), bilinear models (Lee and Elgammal 2004), Hidden Markov Models (He and Debrunner 2000;Kale et al 2004;Wilson and Bobick 1999;Oliver et al 2000), KNearest Neighbor classification (Little and Boyd 1998) or Support Vector Machines (Lee and Grimson 2002) are used to solve the classification problem.…”