The problem if pixel expansion in cryptography is not simple. If is explained, for example, here:
http://eprint.iacr.org/2006/170.pdf[
^].
You will see from the article, that, in the image reconstructed from the
shares (shares and reconstruction are defined in the first paragraph of the article), the contrast always suffers. There are works on optimization of the pixel expansion, but they are hard to find or access. Generally, this is the level of computer science which is typically beyond the capability of 95%+ of those people who are called "software engineers"… :-)
—SA