this has already been done using current and older gen cameras, using a modded firmware you can view the RGB values of each pixel, thus a description that completely describes the image
then using some scripting, you can take the text output of the rgb values, then create an image from it which can be printed and you will have a very accurate visual of what is in the pixel, with perfect pixel mapping (and best of all, no crowd sourcing needed since if the RGB values of each pixel is not enough, then an interpreter can map those values to actual pixels.