Optimum transform coding of imagery

A system is presented for transform coding of imagery. Specifically, the system uses the 2-D discrete cosine transform (DCT) in conjunction with adaptive classification, entropy-constrained trellis-coded quantization, optimal rate allocation, and adaptive arithmetic encoding. Adaptive classification, side rate reduction, and rate allocation strategies are discussed. Entropy-constrained codebooks are designed using a modified version of the generalized Lloyd algorithm. This entropy-constrained DCT-based system is shown to achieve outstanding coding performance as compared to other DCT-based systems.