The core content of the nearest interpolation is very simple. First you need to use the floating point coordinates obtained by inverse transformation and round the floating point coordinates to get the integer coordinates, the pixel value corresponding to this integer coordinate is the pixel value of the target pixel. The pixel value corresponding to the upper-left corner point of floating-point coordinates is the desired pixel value [4]. You can see that the nearest neighbor interpolation is easier to understand and the results are more convenient, but the image quality is not so high.