Raspberry Pi Camera with Steganography features in matlab

Hi guys! I'm working on a final year project, the idea I've come up with is a Raspberry Pi linked with MATLAB, which captures footage using a camera, and then using steganography, some data will be embedded into the picture.

Truth is, I'm an idiot, I don't really know where to start with this, or if it's even possible. I don't even know if it's worth doing a final year project on, if you guys have any other ideas of a project idea with raspberry Pi and matlab, it would be much appreciated!

or if you could suggest where I could start, that would also be a great help :D Thank you!

[Walter Roberson]

It is not clear what the thesis is here? Pi can capture images. It is possible to add steganography. It is not clear whether the steganography is to be added at the Pi or at the host, and it is not clear what is to be done with the video afterwards. No point in just live streaming steganography for display, and there does not seem to be a lot of point in sending the same image every frame. Is the point to be spy-like in secretly embedding data transfer in a live video stream because people would normally expect that you would need a recorded video and bit of a time lapse to have time to embed the data? Is the point to see how high a secret bandwidth you can sustain, given some particular model of Pi?

I just don't know what is being investigated here.

When I read your question and thought for a brief time, I thought of something very different: using the Pi for live path-planning advice for something like skiing or snow boarding or mountain bike racing. I am not sure how difficult of a project that would be to do usefully, but it would be cool and potentially useful if it worked.

The skiing and snow boarding versions would perhaps have lower image complexity but would have to be fast; the mountain bike racing would be at slower speeds but would have much more complex images and a lot more AI would be needed to give useful advice about path -- being able to recognize slope, mud, many different kinds of obstacles, and so on.

To get the speeds up on something like this on a Pi, you would be needing to have installed OpenCV on the Pi and make calls into it. The MATLAB part would probably be most useful in rapid algorithm development, perhaps in training classifiers for deployment to the Pi.