We're using the anchors on the images themselves.

We also alter the text inside of images by breaking it up into layers and then replacing the text layer. There's been a trend to put table of contents directly into images because it looks pretty, including page numbers, on PDFs, but if you convert PDF into something pageless (like Asciidoc), then you need to remove those references and replace them with something like relative placements or a hyperlink to a new anchor.