Need image alone version of depth_image_proc/register

edit

I have multiple sets of RGB-D training data from different camera sources. I would like to use sets together to train a machine learning model. However, I (of course) would like them to be consistent.

depth_image_proc/register is perfect for the translation of the depth image. spot on! I just used the camera params for the first and the destination.

however, I also need to do the exact same thing for the RGB image. Is there existing code/functions that can do this?

Example, if the principal point (center) of the image is off in the first depth image, that is taken into account in presenting the second depth image. I need that same functionality for the rgb image.

Thanks!