a) A compression technique in which groups of picture samples (the vectors) are represented by predetermined codes.

Encoding is done by matching the vectors with code words in a code book, and the addresses of the code book are then sent to the decoder. The picture quality depends widely on suitable code books and the match algorithms.

b) A technique where a vector (usually a square of samples of one color component of an image) are represented by a single number. This number is an index into a code book by which the vector is reconstructed.

The major issues are finding (calculating) a robust code book and how to choose the “best” code book entry for a given input vector.