Loss less Compression Using Joint Predictor for Astronomical Images

Downloading astronomical images through Internet is a slow operation due to their huge size. Although several lossless image coding standards that have good performance have been developed in the past years, none of them are specifically designed for astronomical data. Motivated by this, this paper proposes a lossless coding scheme for astronomical image compressions. We design a joint predictor which combines the. interpolation predictor and partial MMSE predictor. Such strategy benefits from its high compression ratio and low computation complexity. Moreover, the scalable and embedding functions can be further supported. The interpolation predictor is realized by up-sampling the downsampled input image using bi-cubic interpolation, while the partial minimum mean square error (MMSE) predictor predicts the background and foreground (i.e., stars) separately. Finally, we design a simplified Tier-1 coder from JPEG2000 for entropy coding. Our experimental results show that the proposed encoder can achieve higher compression ratio than JPEG2000 and JPEG-LS.