You can find all the published algorithms in section 23.3 here: http://www.visionbib.com/bibliography/contentschar.html#OCR,%20Document%20Analysis%20and%20Character%20Recognition%20Systems
I'm not in the field of watermarking/steganography so I can't make recommendations on what subfields of watermarking might be interesting for you to investigate. But, I'd start off with a description of why people might want to watermark, like branding, detect if their images or videos were "stolen", hidden transmission of messages, etc. Then I'd go into a survey of the different subfields, like visible watermarks, invisible watermarks, video bugs, steganography, etc. Then describe some methods, like replacement, averaging, LSB, Fibonnacci, etc. Describe ways of getting around watermarks, like scaling, rotation, adding noise, cropping, etc. Then describe how some methods are robust to these kinds of attacks, that is, the watermark can still be retrieved despite attempts to alter the image. Then if you're required to invent something new (unlikely for undergraduate students but required for graduate students), pick a method that you think you can improve on and try to code something up.
