Impulse Noise Removal Based on Grey Polynomial Interpolation

In this paper, the one-dimensional grey polynomial interpolation (GPI) presented in is applied to impulse noise removal. The proposed approach is motivated by the following observations. First, a median based noise removal approach can be considered as a variation of neighborhood interpolation which is based on the zero-order polynomial. Second, it is well known that the performance of first-order polynomial interpolation is generally better than that with the zero-order. Therefore, a better quality of restored image can be generally expected when a first-order polynomial interpolation is applied in the noise removal. With the noise detection approach described in, the proposed GPI based noise removal approach is justified by four examples where comparisons are made with the median based approach reported in. The results indicate that the proposed approach generally outperforms the compared approach.