A Nonlinear VQ-based Predictive Lossless Image Coder

A new lossless predictive image coder is introduced and tested. The predictions are made with a nonlinear, vector quantizer based, adaptive predictor. The prediction errors are losslessly compressed with an arithmetic coder that presumes they are Laplacian distributed with variances that are estimated during the prediction process, as in the approach of Howard and Vitter. I. Introduction Predictive coding is a common way of losslessly compressing originally analog data. A block diagram of a typical system is shown in Figure 1. predictor Lossless Encoder Image Prediction Error Prediction Bits Figure 1: A Predictive Lossless Image Coder In such systems, a prediction for each sample to be coded is generated using the sample values previously transmitted to the decoder. This prediction is subtracted from the sample to produce an error value, which is then compressed with a lossless encoder such as a Huffman coder or an arithmetic coder. The decoder can make the same prediction as the coder..