I can't say for sure because I don't have the 3rd party signature handler (gemsignplus) available to see how they are building the signature object, but I'll take my best guess. When the signature is created and added to the PDF file they are probably imbedding the display contents (the captured signature appearance from the Topaz pad) into the signature object. It's not a data object like the text that is is typed into the text form fields, and thus it cannot be extracted.
This gets a bit confusing, because the signature displayed on the page is not really the signature, it's just a pictorial representation. This is provided we're talking about certificate based signing where the actual signature is a cryptographic operation. I've seen some third party signature handlers that only use the captured signature from the signing pad (aka the wet ink signature) and others that also using certificate based signing to provide a check for document integrity.
Still, I think your original question was how do you extract the signature apperance, and that depends on how it's written into the PDF.