A Binary Image Scalable Coder Based on Reversible Cellular Automata Transform and Arithmetic Coding

The paper presents an efficient scalable coding approach for bi-level images that relies on reversible non-linear transformations performed by subclasses of Cellular Automata. At each transformation stage the input image is converted into four subimages which are coded separately. In this work we delineate an effective strategy for the entropy coder to code the transformed image into a binary bit stream that outperforms the compression results previously obtained and compares well with the standard JBIG. Experimental results show that our method proves to be more efficient for images where black pixels lie within a connected region and for multiple decomposition levels.