Gait recognition based on embedded Hidden Markov models

In this paper, a gait silhouette or feature cycle is divided into several temporally adjacent clusters. Each cluster is calculated and gets several adjacent PFDEI binary images. Embedded Hidden Markov mode (EHMM) is built to describe the relationship among the PFDEI. In every PFDEI state we divided several substate to describe the PFDEI image. Frame difference image is good chosen to describe substate. Sparse representation is applied to analyze binary image to learning its parameters.