User-independent myoelectric gesture recognition system based on adaptive learning

Abstract

The invention relates to a user irrelevant myoelectric gesture recognition system based on adaptive learning. The system comprises a data acquisition unit, a clustering unit, an adaptive KNN neighbor classifier and a risk evaluator which are connected in sequence. The data acquisition unit is used for acquiring existing user data and processing the data; the clustering unit finds clustering centers of different actions from the signal data after data processing by adopting K-Means clustering, and extracts N samples with the shortest distance between each action of each user and the clustering center as a training set for training a self-adaptive KNN neighbor classifier; the adaptive KNN neighbor classifier is used for obtaining a corresponding label according to the new user data; and the risk evaluator evaluates new user data, and qualified samples are used for replacing remote samples of the training set and updating weights of the samples of the training set. According to the method, the problem that the model is not universal due to the individual difference of the electromyographic signals is solved, the re-training step of the user is not needed, the use experience of the user is greatly improved, and the recognition accuracy can be dynamically improved.

Description

technical field [0001] The invention relates to the field of gesture recognition, in particular to a user-independent myoelectric gesture recognition system based on adaptive learning. Background technique [0002] With the development of equipment intelligence, the human machine interaction (Human Machine Interaction, HMI) method centered on computers in the past is changing into a natural human machine interaction method centered on people. The natural communication of computers. Since the surface electromyography signal directly reflects the degree of muscle activity and can be analyzed to obtain the user's movement intention, it is widely used in the field of human-computer interaction. [0003] At present, the human-computer interaction system based on surface electromyography usually includes two main processes: 1) use the user's training samples to train the gesture classification model offline; 2) use the trained classification model to perform online gesture recogn...

AI Technical Summary

Problems solved by technology

[0004] In view of this, the object of the present invention is to provide a user-independent EMG gesture recognition system based on adaptive learning, which solves the problem of non-universal models caused by individual differences in EMG signals, and does not require user retraining steps, which is very convenient. The user experience is greatly improved, and the recognition accuracy rate will be dynamically improved

Images