Robust view-invariant multiscale gait recognition

Abstract

The paper proposes a two-phase view-invariant multiscale gait recognition method (VI-MGR) which is robust to variation in clothing and presence of a carried item. In phase 1, VI-MGR uses the entropy of the limb region of a gait energy image (GEI) to determine the matching gallery view of the probe using 2-dimensional principal component analysis and Euclidean distance classifier. In phase 2, the probe subject is compared with the matching view of the gallery subjects using multiscale shape analysis. In this phase, VI-MGR applies Gaussian filter to a GEI to generate a multiscale gait image for gradually highlighting the subject's inner shape characteristics to achieve insensitiveness to boundary shape alterations due to carrying conditions and clothing variation. A weighted random subspace learning based classification is used to exploit the high dimensionality of the feature space for improved identification by avoiding overlearning. Experimental analyses on public datasets demonstrate the efficacy of VI-MGR.

Clothing and carrying condition invariant gait recognition based on rotation forest

“…Hence, the introduction of GEI [11]. Since then many promising model-free gait recognition methods have been proposed based on a GEI,e.g., [26,15,29,24,1,5] to outperform the original method of GEI.…”

“…Fig. 6(b) shows the results of comparisons with GEI and VI-MGR (available from [5]). The figure shows that our method significantly outperforms GEI and VI-MGR at rank-1 CCR.…”

Sparse error gait image: A new representation for gait recognition

“…The performance of the proposed methods is equivalent to the state-of-the-art, with the best performance being achieved by the multiscale method [18], which uses a complex random subspace learning method to perform recognition. It should be noted that the proposed method considers a very simple recognition method, combined with the proposed SEGI representation.…”

“…The Multiscale Method proposed in [18] applies Gaussian filtering to a GEI at different scales. It then selects 500 random principal components, using 2D PCA, followed by 2D LDA.…”

“…This is followed by user recognition performed using Random Subspace Learning. The work is improved in [18] by applying a Gaussian filter to the GEIs at different scales before using them as a feature for recognition. The work presented in [19] performs view detection using a Gaussian process (GP) classifier on a GEI, followed by canonical correlation analysis (CCA) for recognition.…”