how to start a action on text detection??

edit

Hi

OpenCv doesn't contain a OCR (Optical Character Recognition) library. But you can use tesseract it's also open source. A detailed description how tesseract works can be found here.

If your piece of text is always exactly the same, just use a keypoint detector, giving large description regions to the descriptor for each keypoint and then try to find a match between a given and a database image by looking at the amount of matches that are retrieved, resulting in a confidence score.

Information on keypoint features, detectors and descriptors can be found here

Also a good starting point could be to look into the SLAM algorithm, which is explained here