Abstract:Gait-based pedestrian identification has important applications in intelligent surveillance. From anatomical viewpoint, the physical uniqueness of human gait is physiological discriminative of individuals. Therefore, in theory, like fingerprint and face, gait is used as a biometric for pedestrian identification. However, gait-based pedestrian identification still faces multiple challenges due to a vast diversity of walking conditions and complex acquisition environment during data collection. In this paper, through combining the nonlinear dimensionality reduction by using gait manifold and the temporal feature of gated recurrent unit (GRU) together, we propose a novel gait-based pedestrian identification framework. Firstly, we design a temporal enhancement module to construct a series of frame-by-frame gait trend energy images (ff-GTEIs), which represents spatiotemporal gait characteristics and does not reduce the number of samples. Secondly, a supervised locally linear embedding (LLE) dimensionality reduction scheme is proposed, which generates a low-dimensional gait manifold for each pedestrian and transforms all ff-GTEIs into corresponding gait manifold space. Thirdly, a new pedestrian identification network based on residual GRU is proposed, which is able to identify a person by comprehensively considering the similarity between its gait and corresponding gait manifold. Finally, a series of comparative experiments are carried out based on well-known gait datasets, the experimental results show that the proposed framework for pedestrian recognition in this paper exceeds most existing methods, and has achieved an average correct recognition rate 97.4% and 99.6% based on the open-accessed gait dataset CASIA Dataset B and OU-ISIR LP dataset, respectively.

A Gait Recognition Algorithm Based on Angle Histogram

Hybrid Silhouette-skeleton Body Representation for Gait Recognition

Accelerometer-based Gait Recognition Via Voting by Signature Points

A Novel Algorithm of Gait Recognition

Human Gait Recognition Based on Self-Adaptive Hidden Markov Model

Non-local Gait Feature Extraction and Human Identification

Cross-view Gait Recognition Through Ensemble Learning

Gait recognition based on multiple image silhouette feature combination

Gait Recognition Based on Fast Shape Matching

Gait Recognition Method for Arbitrary Straight Walking Paths Using Appearance Conversion Machine

Automated Gait Recognition Using Weighted DTW Distance

Human Identification Based on Gait Manifold

Human Identification Based on Gait Sequences

Human Gait Recognition by Integrating Motion Feature and Shape Feature

Automatic gait recognition from a distance

Gait recognition via independent component analysis based on support vector machine and neural network

Human Gait Recognition Based on Frame-by-Frame Gait Energy Images and Convolutional Long Short-Term Memory

Gait recognition using wavelet descriptors and independent component analysis

Gait Recognition Via Flexible Measurement

Wearable Device-Based Gait Recognition Using Angle Embedded Gait Dynamic Images and a Convolutional Neural Network

Gait Recognition Using Procrustes Shape Analysis And Shape Context