I ve written some many years ago dynamic Huffman algorithm to compress and decompress data. It is mainly targeted to data with some symbols occuring more often than the rest (e.g. having some data file consisted of 3 different symbols and their total number of occurence in that file s1(1000), s2(200), s3(30) so the total length of file is 1000+200+30=1230 bytes, it will be encoded assigning one bit to s1 and 2 bits to s2, s3 so the encoded length will be 1*1000+2*(200+30)=1460 bits=182 bytes). In the best case the file consisted of just one symbol will be encoded with compression ratio as 1:8. Huffman coding is used in image compression, however in JPEG2000 arithmetic codec is imployed.