Gait Recognition Via Coalitional Game-based Feature Selection and Extreme Learning Machine

摘要

In order to achieve the goal of controlling the intelligent lower limb prosthesis effectively, it is very crucial to recognize the gait pattern of the lower limb, which usually includes walk, up and down stairs or slopes, etc. This paper proposes a gait recognition method based on coalitional game-based feature selection and extreme learning machine. Firstly, this paper extracts characteristic values of four periods in gait cycle, obtaining 24 features. Secondly, in order to improve the accuracy and reduce the computational complexity, a coalitional game-based feature selection algorithm is used to select the prominent features. Lastly, the extreme learning machine (ELM) is used to recognize the gait pattern, which can have a better result in identifying the five kinds of gait pattern in this experiment, compared with BP neural network. Compared with other feature selection algorithms, including mRMR and Relief-F, the proposed method selects fewer features and provides higher accuracy and has faster recognition speed, which proves the effectiveness and feasibility of the proposed method.