Image Geometry Transforms#

This chapter describes implementation of the oneIPL image processing functions that perform geometric operations of image resizing, rotating, warping and remapping.

Most functions performing geometric transform of an image use an interpolation algorithm to resample the image. The type of interpolation method to be used is passed to the function in the interpolation parameter for rotate, warp, and remap. For resize transform, the interpolation type is part of the function name.

The following interpolation algorithms are used:

  • bilinear interpolation

  • bicubic interpolation

  • supersampling

  • interpolation using Lanczos window function

The nearest neighbor algorithm is the fastest, while other methods yield higher quality results, but are slower.


Implementation may use interpolation with edge smoothing option only in those geometric transform functions where this option is explicitly listed in the parameters definition section.

Supersampling mode of resize transform has several limitations. It shall be used only:

  • for reducing image size

  • for full images resize, while other interpolation modes can be used for full and tiled images for better speed/memory performance.