The above suggestions are good if your images always contain text (must be OCRed). A more general application would be to use a library such as OpenCV - it gives you several ways to extract objects (edges, shapes, colors, etc.) from images and compare them.
An even simpler method is to use the OpenCV pattern matching method, which βcomparesβ one image (template) with another.
Since you are in C #, you should look at the Emgu.NET shell for OpenCV .
Mikos source share