Micrographs for single particle analysis would not normally be collected with any annotations (scale bar, micrograph number, etc), unless you're collecting on film and using a scanner (which I doubt anyone does any more?) What type of microscope was used, and what sort of camera/detector does it have?
In the end, all of the data used for single particle reconstruction must have the same magnification. Data for single particle reconstruction should all be collected at a single magnification, ideally on the same microscope in a similar period of time (since microscope magnification can drift by ~1% over long periods of time).
If you're stuck with images at different magnifications you have to decide which ones you will actually use, understanding that your final resolution will be limited to that of the lowest magnification image. That is, if you have images at 2 A/pix, 2.7 A/pix and 4 A/pix you would need to rescale all of the images to 4 A/pix (using e2proc2d.py --fouriershrink). If you opt to not use the 4 A/pix images, you would downsample the 2 A/pix images to 2.7 A/pix.
Upscaling (increasing the effective magnification) does not improve the resolution of the images, it just makes the pixels bigger, and including such images will generally degrade the quality of the final structure rather than helping.
Also keep in mind that due to stain penetration issues, generally speaking the reconstructed envelope will be limited to ~20-25 Å "resolution", though people have demonstrated cases where they were able to achieve subnanometer resolution in specific systems with extremely careful staining processes. It's been a long time since I've looked at any of this, so my memory of the limitations of such methods has faded a bit. Note also that the typical GSFSC measure of resolution used for CryoEM doesn't really give valid numbers in negative stain. You can often achieve "resolutions" much better than 20-25 Å, but this is just saying that the edge of the stain envelope is sharper, not that you will see finer details in the structure.