Personal Identification Methods Using Footsteps of One Step

In recent years, it has been shown that it is possible to identify individuals by walking footsteps, and research is being conducted to improve the recognition rate. The conventional method targets multiple footsteps and imposes many restrictions on the recording conditions. The task has further required precise, manual segmentation of the sequence of footsteps for analysis of the observation waveform. On the other hand, in our previous research, we have proposed a method that can robustly identify only the footstep of one step, with few restrictions for recording conditions and extraction methods. In this paper, we investigate whether our proposed method can identify individuals only by the difference in walking sound for each person regardless of the presence and difference of shoes. As a result, we found that the proposal method using SVM or CNN can identify with average accuracy 98 % and distinguish between subjects without being affected by shoes.