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.

Gabor Phase Embedding Of Gait Energy Image For Identity Recognition

Gabor Phase Based Gait Recognition

Gabor-Based Discriminative Common Vectors for Gait Recognition

Manifold Learning Based Gait Recognition Using Gabor Phase Spectrum

Fast communication: Gait recognition based on dynamic region analysis

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

Individual recognition using gait energy image.

Gait recognition based on structural gait energy image

Gabor-based DCV gait recognition using enhanced gait energy image

Human Identification Based on Gait Manifold

Non-local Gait Feature Extraction and Human Identification

Gait Recognition Based on Gabor Wavelet and Local Binary Pattern

New Gait Recognition Method Based on Static and Dynamic Features

Enhanced gait recognition based on weighted dynamic feature

Gait Recognition Via GEI Subspace Projections and Collaborative Representation Classification

Human Gait Analysis using Gait Energy Image

Speed-Invariant Gait Recognition Using Single-Support Gait Energy Image.

Gait Recognition Using Flow Histogram Energy Image

Gait Recognition Method Based on Dynamic Energy Feature

Person Re-identification by analyzing Dynamic Variations in Gait Sequences

Human Gait Recognition Based on Self-Adaptive Hidden Markov Model