transform_img clips image, doesn't have option to specify output size

[bsugerman]

I noticed that transform_img clips the output image to the same size as the input image, and it would be very useful if one could specify an option for an output-image size or a suitable flag so the entire input image was shown after applying the transformation. For example, I need to find the shift and rotations to geometrically register images, but I am then scaling them up (zooming in) by some factor and combining them together to produce an enhanced, larger image (this is the initial step for super-resolution).

skimage allows this with e.g. warp(). Could you consider adding such a feature?

[SebastianHD]

I'm also surprised that this does not exists. The transform_img function does return "The transformed img, may have another i.e. (bigger) shape than the source" which indicates that this is a possibility, but there seems to be no flag to control it. Right now I'm resorting to first creating a padded image to give to tranform_img function.

[matejak]

This is actually the right approach. The constant image dimensions are important wrt the DFT, so in order to prevent unwanted clipping, one has to have a buffer made of padding. I think that making images twice as large is safe enough. A big difference in scale is not something that DFT will cope with easily.