Has anyone seen research about using the entire Q vocabulary to build a dictionary and attempt brute forcing Q's pic drops for text hidden behind steganography?

Eg.

1. Build dictionary.

1. Pick word, decode image.

1. Decoded image* contains text? If so: quit.

1. Else: Goto 2.

*Challenging, but possible. Open source libs for OCR are good.