In the image compression system, the compression of images is treated as an optimization problem. The goal is to encode the image using a minimum of bits that provide a minimum distortion in the reconstructed image. Each stage of the image compression system may be optimized independently, which often leads to good, yet overall suboptimal results. For example, Coifman and Wickerhauser [CW92] proposed an efficient scheme for selection of the best wavelet packet basis by minimizing information cost. However, an additional procedure would be needed to find the optimal encoding within that basis given a total bit constraint. The simultaneous optimization over multiple stages can attain better results. For example, Ramchandran and Vetterli's algorithm, which simultaneously finds the best wavelet packet basis and quantizers, shows more favorable results [RV93].