While yes, as Walter says, you potentially could write a converter, it seems to be the opposite workflow to what people normally do. So, what would be the reason to want to do this?
There are a few technical considerations to bear in mind as well:
- your image is probably 8-bit per pixel. The RAW10 is 10-bit per pixel. So you'll always have the two highest bits wasted. If the image was initially RAW10, you're never going to recreate the additional 2 bits of information that were lost.
- RAW10, like all raw formats assumes the decoder knows what sensor the image corresponds to. Therefore, the RAW10 format does not include any information about the image such as width and height and bayer pattern. The image is just one long line of pixels. It's up to the decoder reshape that into a rectangle of the right size and to remove potential padding pixels (if the number of columns is not a multiple of 4).
- If the image is colour, then the pixels in RAW10 needs to be encoded using a bayer pattern. There are several possible bayer patterns (GBRG, GRBG, BGGR, RGGB) so you'd have to chose the one that matches your sensor/decoding software. An additional problem is that while matlab has a function to decode bayer patterns it does not have one for encoding into them, so you'd have to write your own.
Once the image has been bayer encoded it's easy to convert it to RAW10. I can write the code for that if really needed (but not the bayer encoding).
