Complexity histograms of a non-compressed and a compressed file


The figures below illustrate the "file-complexity histograms." Original files (that are non-compressed) have variety of histogram patterns (black curves), while the compressed files (by ZIP or LHA) tend to converge in a "normal distribution" pattern (red curve). It is known that a "random file's" histogram is approximated by a Normal Distribution having MeanVal=0.5, Stand. Dev=0.047. The compressed JPEG file below shows almost a random file pattern.


a TEXT file


a Word file


a PDF file


a JPEG file



An embedding file (a file to be embedded) is compressed before embedding operation in BPCS-Steganography. The embedding operation starts from the LSB-plane and goes up to upper-planes in each color. Therefore, if the embedding operation stays only in a small amount, the image degradation after embedding will never be detectable by any means.


(Updated on June 6, 2008 by Eiji Kawaguchi)