We now turn to the more general case illustrated by the “Code 3” column in Table 1.1 on page 7. It is the best of the three listed codes because, somehow, its set of codeword lengths better matches the probability distribution than do the other two sets. Which forces the question: given a sorted list of symbol probabilities, how can a set of prefix-free codewords be assigned that is best for that data? And what is really meant by “best”?
KeywordsCode Algorithm Huffman Code Symbol Number Codeword Length Prefix Code
Unable to display preview. Download preview PDF.