The functions described in this section operate with universal image pyramids. These pyramids use separable symmetric kernel (not only Gaussian type) and downsampling/upsampling with arbitrary factor (not only 2). The next pyramid layer can be built for an image of an arbitrary size. These pyramids are used in some computer vision algorithms, for example, in optical flow calculations.
All universal pyramid functions use the mirrored border.
Example shows how to build pyramids and calculate the optical flow for two images.