Improving single-pass adaptive VQ

Constantinescu and Storer (1994) introduced a single-pass vector quantization algorithm that with no specific training or prior knowledge of the data was able to achieve better compression results with respect to the JPEG standard, along with a number of computational advances such as: adjustable fidelity/compression tradeoff, precise guarantees on any l×l sub-block of the image, and fast table-lookup decoding. We improve that basic algorithm by blending it with the mean shape-gain vector quantization (MSGVQ) compression scheme. This blending allows a slightly better performance in terms of compression and a clear improvement in visual quality