One-Dimensional Grey Polynomial Interpolators for Image Enlargement

In this paper, one-dimensional (1-D) grey polynomial interpolators (GPIs) for image enlargement are proposed. Note that (i) the randomness inherent in image data affects the performance of polynomial interpolators (Pis) in image enlargement, and (ii) that the preprocessing scheme in grey systems, the first-order accumulated generating operation (I-AGO), is able to reduce randomness in data. In this paper, 1-D grey polynomial interpolators are developed for image enlargement where 1-AGO is used to preprocess image data. To improve the performance of GPIs further, an a filter is applied to smoothen the interpolated pixels. Examples are given to verify the proposed 1-D GPIs. Simulation results indicate that the 1-D GPIs have better performance than the 1-D Pis in terms of PSNR. Besides, the 1-D GPIs are compared with well-known 2-D polynomial interpolators in image enlargement, i.e., the nearest-neighbor interpolator, the bilinear interpolator, and the bicubic interpolator. Interesting enough, the 1-D 2-order GPIs performs little better than the 2-D bicubic interpolator in PSNR on average, with equally well visual quality.