A Signal Denoising Method of Gesture Radar Based on Weighted Principal Component Analysis and Improved Wavelet Threshold

When using radar to recognize gesture, radar echo signals are susceptible to noise. In order to improve detection resolution and data interpretation quality of gesture radar, this paper proposes a new denoising algorithm based on weighted principal component analysis and improved wavelet threshold, named WPCA-IWT denoising algorithm. First, radar echo data are standardized and each echo data’s variance are calculated. Each echo data is multiplied by it’s corresponding variance to complete weighting operation. By using WPCA processing, the dimension of echo data is reduced and echo signal’s principal components are extracted. Then, a new threshold function is proposed and an improved wavelet threshold denoising algorithm is designed to denoise principal components. Finally, denoised principal components are multiplied with an eigenvector matrix to reconstruct radar echo. The proposed WPCA-IWT denoising algorithm has obvious advantages at the aspects of efficiency and denoising effect. Both simulation and on-site radar data experiments verified the effectiveness of the proposed algorithm.