Smart compression using the embedded zerotree wavelet (EZW) algorithm

The embedded zerotree wavelet (EZW) algorithm has proven to be an extremely efficient and flexible compression algorithm for low bit rate image coding. The embedding algorithm attempts to order the bits in the bit stream in numerical importance and thus, a given code contains all lower rate encodings of the same algorithm. Thus, precise bit rate control is achievable and a target rate or distortion metric can be met exactly. Additional, the technique is fully image adaptive, does not even need to know the dynamic range of the numbers being compressed. Because of the embedding and adaptivity features of EZW, judicious pre-processing at the encoder and the corresponding inverse post-processing at the decoder can be used to convert the numerical optimization used in EZW to other criteria such as a subjective one or one which favors critical regions as illustrated by an application called "smart compression".<>