Human Gait Identification Using Persistent Homology
持久同源性在人类步态识别中的应用
Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications, Volume 7441 of the series Lecture Notes in Computer Science pp 244-251
Abstract: This paper shows an image/video application using topological invariants for human gait recognition. Using a background subtraction approach, a stack of silhouettes is extracted from a subsequence and glued through their gravity centers, forming a 3D digital image I. From this 3D representation, the border simplicial complex ∂ K(I) is obtained. We order the triangles of ∂ K(I) obtaining a sequence of subcomplexes of ∂ K(I). The corresponding filtration F captures relations among the parts of the human body when walking. Finally, a topological gait signature is extracted from the persistence barcode according to F. In this work we obtain 98.5% correct classification rates on CASIA-B database.
Gait-Based Person Identification Using Motion Interchange Patterns
运动交换模式在基于步态身份识别中的应用
Computer Vision - ECCV 2014 Workshops, Volume 8926 of the series Lecture Notes in Computer Science pp 94-110, Date: 20 March 2015
Abstract：Understanding human motion in unconstrained 2D videos has been a central theme in Computer Vision research, and over the years many attempts have been made to design effective representations of video content. In this paper, we apply to gait recognition the Motion Interchange Patterns (MIP) framework, a 3D extension of the LBP descriptors to videos that was successfully employed in action recognition. This effective framework encodes motion by capturing local changes in motion directions. Our scheme does not rely on silhouettes commonly used in gait recognition, and benefits from the capability of MIP encoding to model real world videos. We empirically demonstrate the effectiveness of this modeling of human motion on several challenging gait recognition datasets.
An Approach to Emotion Identification Using Human Gait
基于人类步态的情感识别方法
Proceedings of Fourth International Conference on Soft Computing for Problem Solving, Volume 336 of the series Advances in Intelligent Systems and Computing pp 165-175, Date: 24 December 2014
Abstract: Human gait data have abundant information for recognition of actions, intentions, emotions, and gender. The paper presents an approach toward classification of human emotions using gait data into three classes: happy, angry, and normal. Data of human gait for 3 emotional expressions (happy, angry, and neutral) of 25 individuals were collected. The silhouette was divided into 9 segments in order to analyze motion in various body parts moving with different frequency. The features such as centroid, aspect ratio, and orientation were extracted from different segments using geometric and Krawtchouk moments, respectively. A train model was generated from testing data using support vector machines (SVM), and hence, new feature vector was classified into three classes. The results show that polynomial kernel using geometric moment features has the maximum recognition rate of 83.06 %.
Human biometric identification through integration of footprint and gait
脚印与步态之整合技术在人类生物特征识别中的应用
Intelligent and Information Systems, International Journal of Control, Automation and Systems, August 2013, Volume 11, Issue 4, pp 826-833
Abstract: Gait recognition has gained attention from the biometric community because it has a couple of advantages over other biometric methods to identify individual humans: (1) it requires no subject contact and (2) gait can be assessed from a distance when other physical measures might be obscured or not available. However, objects carried or worn by a subject, notably a briefcase or overcoat, may deform the gait silhouette and significantly degrade the performance of the gait recognition system. In this paper we propose that footprint and gait information may be combined to create a new method for human identification. This method automatically partitions the gait cycle based on the footprint and fuses these two parameters at the decision level to improve accuracy. We have applied the proposed algorithm to a USF gait data set to demonstrate its performance.
Gait Biometrics: An Approach to Speed Invariant Human Gait Analysis for Person Identification
步态生物特征：速度不变人类步态分析法在身份识别中的应用
Proceedings of the Second International Conference on Soft Computing for Problem Solving (SocProS 2012), December 28-30, 2012,Volume 236 of the series Advances in Intelligent Systems and Computing pp 729-737
Abstract: A simple and a common human gait can provide an interesting behavioral biometric feature for robust human identification. The human gait data can be obtained without the subject’s knowledge through remote video imaging of people walking. In this paper we apply a computer vision-based technique to identify a person at various walking speeds, varying from 2 km/hr to 10 km/hr. We attempt to construct a speed invariance human gait classifier. Gait signatures are derived from the sequence of silhouette frames at different gait speeds. The OU-ISIR Treadmill Gait Databases has been used. We apply a dynamic edge orientation histogram on silhouette images at different speeds, as feature vector for classification. This orientation histogram offers the advantage of accumulating translation and orientation invariant gait signatures. This leads to a choice of the best features for gait classification. A statistical technique based on Naïve Bayesian approach has been applied to classify the same person at different gait speeds. The classifier performance has been evaluated by estimating the maximum likelihood of occurrences of the subject.
Multi-view Gait Fusion for Large Scale Human Identification in Surveillance Videos
多视角步态融合在监视视频大规模身份识别中的应用
Advanced Concepts for Intelligent Vision Systems, Volume 7517 of the series Lecture Notes in Computer Science pp 527-537
Abstract: In this paper we propose a novel multi-view feature fusion of gait biometric information in surveillance videos for large scale human identification. The experimental evaluation on low resolution surveillance video images from a publicly available database [1] showed that the combined LDA-MLP technique turns out to be a powerful method for capturing identity specific information from walking gait patterns. The multi-view fusion at feature level allows complementarity of multiple camera views in surveillance scenarios to be exploited for improvement of identity recognition performance.
Feature Extraction and HMM-Based Classification of Gait Video Sequences for the Purpose of Human Identification
步态视频序列特征提取及基于HMM分类在身份识别中的应用
Vision Based Systemsfor UAV Applications, Volume 481 of the series Studies in Computational Intelligence pp 233-245
Abstract: The authors present results of the research on human recognition based on the video gait sequences from the CASIA Gait Database. Both linear (principal component analysis; PCA) and non-linear (isometric features mapping; Isomap and locally linear embedding; LLE) methods were applied in order to reduce data dimensionality, whereas a concept of hidden Markov model (HMM) was used for the purpose of data classification. The results of the conducted experiments formed the main subject of analysis of classification accuracy expressed by means of the Correct Classification Rate (CCR).
