A New Gait Recognition Method Based on Body Contour

The reliable extraction of gait features from image sequences and their recognition are two important issues in gait recognition. An appearance-based approach to the problem of gait recognition is proposed in this paper. The image feature templates are represented by the vectors obtained by horizon, vertical and diagonal scanning of the outer contour of binarized silhouette of a walking person. A subspace transformation, which combines principal component analysis (PCA) with linear discriminant analysis (LDA), is applied to process the spatial templates. It aims essentially at reducing data dimensionalities, and optimizing the class separability of different gait sequences. This new approach is applied to a mixed data-set. Extensive experimental results demonstrate that the proposed algorithm performs an encouraging recognition rate of 87% with relatively lower computational cost