I have a random binary image of size RxC (R>=256 and C>=256). There is no any kind of adjacency property at all in the image. I need to compress the same, (lossless compression) that need to be represent with less than R*C number of bits, including all the overheads. Any suggestion on this regard will be very much helpful. Thanks in advance.
If you are looking for algorithm strictly dedicated to binary images check the method introduced in Lossless binary image compression using logic functions and spectra by B. J. Falkowski.
Article Lossless binary image compression using logic functions and spectra
I'm no expert in this field, but my gut feeling is that if the pixels are random 0 or 1, you need as many bits as there are pixels, plus the overhead specifying the dimensions of the image (4 bytes or so).
I have always thought that random bits don't compress well (at least in the JPEG codec).
Tilo Strutz I know I am too late to acknowledge your answer. Thanks for your answer.
You can always apply Run Length Encoding (RLE) if there are large number of repeated values of 0 or 1. Moreover if there are specific patterns of 0 and 1 occurring frequently, then you can use Lempel-Ziv (LZ) coding or Huffman coding.
- Badges
- Science topic