Resampling is a nonparametric method of statistical inference that extracts repeated samples from the original data sample, that is, does not use a common distribution to approximate the value of the calculated probability P.

The resampling method generates a unique sampling distribution based on actual data. It is generated using empirical methods rather than analytical methods. It can be understood as obtaining unbiased estimates based on unbiased samples of all possible outcomes of the data.

Commonly used resampling methods

• Nearest Neighbor ResamplingAccording to the width (height) ratio of the target image to the original image, the pixel points at the relative position of the original image are used as the pixel points of the target image;
• Bilinear Resampling:Refer to the values of the four points around the corresponding position of the original pixel, and take the corresponding weights according to the relative positions to obtain the target image;
• Bicubic Resampling:Refer to the values of 4 * 4 pixels around the original pixel and use them to obtain the target image;
• Lanczos Resampling:The special form of the Arnoldi algorithm for symmetric matrices can be used in the Krylov subspace method and eigenvalue problem for solving symmetric matrix linear equations. This algorithm refers to more original image pixel values, and the amount of calculation increases, but the effect is also the best.