Main Article Content

Abstract

The myoelectric signal is very popular in robotic rehabilitation due to its high success rate and robustness for controlling the robotic device. For controlling any biomedical signal inspired external robotic devices, classification accuracy plays a pivotal role. This paper presents the classification of elbow and four fingers movement by using K-Nearest Neighbors (K-NN) algorithms with Discrete Wavelet Transform (DWT) and Principle Component Analysis (PCA). In conjunction to this, DWT is utilised for feature extraction and de-noising purpose whereas PCA is used for reducing the size of data before the classification. In DWT, Daubechies 4 (db4) wavelet filter is employed and forth level approximation coefficient of reconstructed signal (a4) is used for time scale feature extraction. The result shows that Fine K-NN has 95.67% accuracy which is best among the other K-NN algorithms with selected frequency domain, time domain, and time-frequency domain (time scale) features.

Article Details