Selective Application of Burrows-Wheeler Transformation for Enhancement of JPEG Entropy Coding

JPEG encodes a still image through a lossy encoding followed by a lossless entropy encoding. The entropy coding in JPEG is a combination of the runlength coding and a variable-length coding. The entropy encoding employed in JPEG aims to exploit repetition of strings within a block but not across blocks. We propose application of Burrows-Wheeler transformation (BWT) to exploit repetition of strings across blocks in the entropy encoding. We measure the degree of repetition (DOR) of a block of data for the entropy encoding first and apply BWT to the block if DOR is greater than a preset threshold value. Experimental results show that the proposed method improves the compression ratio, on average, by 18% compared with that of the original JPEG algorithm. Index terms – JPEG, Image processing, Image compression, Lossless compression, Burrows-Wheeler transformation, Entropy encoding