Footstep classification using wavelet decomposition

The characteristics of human footsteps are determined by the gait, the footwear and the floor. Accurate footstep analysis would be useful in various applications, home security service, surveillance and understanding of human action since the gait expresses personality, age and gender. The feasibility of personal identification has been confirmed by using the feature parameter of footsteps, however, it is necessary to use more effective parameters since the recognition rate of this method decreases as the number of subjects increases. In audio classification, Fourier and wavelet transform were used to extract the feature of audio signals. The feasibility of a footstep classification using Fourier and wavelet parameters were confirmed previously. In this paper, we focused on the wavelet parameter which consists of subband power, time-brightness and time-width. Previous work shows that the feature extraction using wavelet transform is effective for footstep categorizations, however, an optimal frame length for feature extraction and the relationship between a recognition rate and the length of feature parameters are not discussed in that paper. This paper provides two dominant results; the frame window size, which yield the good accuracy for footstep classification, is 4096; the feature parameter based on wavelet parameters can be reduced to 2/3 with equivalent recognition rate. Results show that the parameter applied herein yields effective and practical footstep classification.