You can make use of the parser package of iText (Sharp) to find the position of a given text. You do have to implement your own RenderListener, though, as the main use case of that package is text extraction, not text position finding.
It is not as easy as you might think as e.g. the individual characters of the words might come in separately in any order.
PS:
First you will have to find out, though, whether the line for the signature consists of characters (as your question seems to imply) or whether it is a drawn path. Additionally you will have to find out whether that line is unique in the document.
In the former case, the RenderListener implementation you need has to inspect the TextRenderInfo objects forwarded for processing in its RenderText method. If its text content contains those unique characters building the signatrue line, you have to store the position data of this TextRenderInfo. If the line characters are not unique, you will have to find some additional criteria making them unique, e.g. some preceding string or possibly a fact that its the last occurance of those characters in the document.
In the latter case the parser package functionality has to be somewhat extended as it currently does not report paths. According to the iText mailing list, an extension like that is on the ToDo list.
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…