Application of EMG Signal Processing and Classification Based on KNN in the Field of Gesture Recognition

This paper achieves the recognition and classification of electromyographic (EMG) signals for various hand gestures using advanced digital signal processing and machine learning techniques. The raw EMG data was converted into a structured CSV format, optimizing the extraction of categorical labels and enabling comprehensive data visualization. Features of the analyzed signals were extracted using the Fast Fourier Transform (FFT), while denoising with wavelet transform significantly improved signal fidelity and quality. By leveraging the K-Nearest Neighbors (KNN) algorithm, the paper achieved reliable gesture classification, accurately identifying actions such as “hand at rest” and “wrist flexion.” With an average classification accuracy of 96.93 %, the system demonstrates robust potential for integration into realtime motion monitoring and human-computer interaction applications. These findings underscore the effectiveness of EMG-based recognition systems and suggest promising avenues for future research in refining classification performance and practical adaptability.