The use of transform coding for blocks of pels has proven to be an effective way of compacting the block energy, namely the information, in a few transform coefficients. Not all of these need to be transmitted to maintain good image quality, and they need not be represented with full accuracy [NH95]. The most effective with respect to compression performance for images is the Karhunen-Loeve Transform (KLT). However, its high computational complexity has led to the adoption of the Discrete Cosine Transform (DCT) for the majority of applications.