Scalable object-based compression algorithm for segmented space-telescope images
Helen H. Boussalis, Charles Yu Liu, Khosrow Rad, Jianyu Dong · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2004
The noise-alike nature of astronomical images imposes a great challenge on compression. Due to the lack of correlation among adjacent pixels, it is very difficult to achieve good compression result using standard algorithms. To address the above challenge, a novel object-based compression method is proposed in this paper. Based on object analysis, the astronomical entities presented in the image are classified into two categories: clear and faint objects. For the former, a zerotree based wavelet compression algorithm is employed to achieve scalable coding; for the latter, a predictive coding method is used to preserve their location and intensity. The objective is to enhance the detection of faint object in astronomical images while providing a good overall visual effect. Experiment results demonstrate the superior performance of our proposed algorithm.