Actually, MobileFrontend interprets all captchas to be an image. ConfirmEdit supports other modules, too, which are used especially on third party wikis, which ask questions or something similar. For this, the editor should support plain text CAPTCHAs, too.
Is it possible to support <img> tag, display image?
For example, one of our question is: "What's her name: <img src="http://static.mengniang.org/common/thumb/8/83/25208068.png/250px-25208068.png" alt="" title="" />"
I don't really understand the question. I would hope the captcha API would return everything needed for the client to construct the interface
question: "What's her name?",
Passing an answer to the api with questionid 1 would then either let editing complete or not.