Linear-Prediction-Based Multi-resolution Approach for Lossless Image Compression

We design a linear-prediction-based multi-resolution approach for lossless image compression. The linear prediction technique computes the weighted differences between six neighboring pixel values to estimate the predicted pixel value. The prediction error is decomposed by a one-level integer wavelet transform to improve the prediction. The performance of the proposed approach is compared with the lossless Joint Photographic Experts Group (JPEG) and lossless adaptive linear predictor schemes. Our approach is compared with these two schemes because these techniques estimate the present pixel value from the previous pixel values. Our proposed scheme yields lower bits/pixel (higher compression) than the lossless JPEG, which is a purely linear-prediction-based approach. It also yields a comparable or lower bits/pixel than the lossless adaptive linear predictor scheme. However, the computational complexity is greatly reduced because the predictor coefficients are known by both the encoder (compressor) and the decoder (decompressor). This advantage is particularly attractive in real time processing for compressing and decompressing digital images.