Human gait feature extraction method

In this paper, a new method is proposed for human gait feature extraction from a video. The method consists of seven stages: background/foreground segmentation; noise filtration; extracting of the human silhouette; dividing the human silhouette into eight horizontal segments based on human body proportions; bounding rectangles getting; phase synchronization; features calculation; Fourier transform (optionally). A support vector machine (SVM) algorithm is used for classification. The algorithm was tested on 102 gait samples. Recognition accuracy is 89.2% 96.5%. (C) 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC -ND license (hapslicreat0 ecommons.orgincensesiby-0040114 0) Peer-review under responsibility of the scientific committee of the 10th International Young Scientists Conference on Computational Science

Automated summary

A new method for human gait feature extraction from videos is proposed in this paper, utilizing seven stages and support vector machine (SVM) algorithm for classification, with a recognition accuracy of 89.2% to 96.5% in experiments.