Lossless Compression of MODIS Data Based on the Maximum Spanning Tree and 3D Context Prediction

MODIS data is increasingly important for oceanographic, terrestrial, and atmospheric science observation. Because of the high data rate, the loss less data compression becomes vital for MODIS data transmission and storage. In this paper we present a new approach for loss less compression of MODIS data based on the maximum spanning tree and 3D context prediction. First we determine the prediction sequence using the maximum spanning tree derived from the inter-band correlation coefficient matrix, then the three dimensional context prediction method is performed. At last the reference band and residual bands are compressed using JPEG-LS. Experimental results show that our method outperforms WinRAR and JPEG-LS.