Multi-pattern recognition of the forearm movement based on SEMG

In the article, a new feature extraction method of surface electromyography (SEMG) is introduced. It is called power spectrum coefficient method. This method defines the ratio of maximum energy-band spectrum and power spectrum as an eigenvalue, mostly depressing the influence of special person. By using Bayes statistics algorithm in the power spectrum coefficient method, multipattern recognition of the forearm movement is fulfilled. The experiment verified that it is effective for recognition, and in the state of nonspecific-person, the correctness of recognition reaches eighty-four percents.