Abnormal Gait Recognition based on Integrated Gait Features in Machine Learning

Human gait has various movements by individuals and enables normal and abnormal gaits to be recognized by their characteristics. Gait abnormalities are the most common symptoms caused by various neurological disorders such as hemiparetic, myopathic, sciatic neuralgia, and Parkinson’s disease. Abnormal gaits from functional disorders have different types of characteristics. Based on the characteristics, we propose a gait abnormality recognition method that uses integrated gait features extracted from the individual’s walking movement using the Kinect depth camera. Diagnosing a detailed pathological level of the disorders requires a high knowledge of the physiological and pathological gait characteristics. Furthermore, it commonly requires a much complex environment and wearable sensors. However, using our proposed method, the initial characteristics of the gait disorders can be detected without the requirements. Thus, we build a k-NN classifier and an SVM classifier to classify abnormal gait from a walking person. In the result of our experiment, our proposed method with the gait features shows the potentiality for classifying gait abnormalities.