2-D Median Filter-Based Impulsive Noise Reduction in Multifrequency Phase-sensitive Demodulation of Electrical Impedance Tomography

Abstract

Multifrequency phase-sensitive demodulation (PSD) plays an important role in electrical impedance tomography (EIT). In the presence of Gaussian noise, multifrequency PSD can be regarded as a matched filter and an optimal linear filter for the acquired signal. However, the additive noise often has the characteristics of impulsive noise in a real measurement system. Thus, it is a challenging task to reduce the impulsive noise effectively in multifrequency PSD. Considering the periodicity of multifrequency signal and the significant properties of 2-D median filter (MF), a method for impulsive noise reduction in multifrequency PSD of EIT was proposed in this paper. This paper mainly focused on reducing the impulsive noise by using the characteristics of signal itself and the 2-D MF. Numerical simulations and experiments were set up to verify the proposed method. Both results demonstrated that the proposed method can reduce the impulsive noise in multifrequency PSD effectively, and it can also provide demodulation information with higher signal-to-noise ratio and stronger robustness.