FAST ALGORITHM FOR REMOTE SENSING IMAGE PROGRESSIVE COMPRESSION

A new fast algorithm for the remote sensing image progressive compression was proposed. This algorithm has three embedded characters (resolution, region of interest, and fidelity), low computing complexity and favorable loss compression performance. Every resolution of wavelet transform coefficients were partitioned into many precincts according to the area. In each sub-band of each precinct, the spatio-temporal neighborhood relationship was used to remove redundancies between different bit-planes and neighbors in the same bit-plane, and the bits of every bit-plane were modeled and reordered to form three sub-processes and run-length encoded only in one pass. The adaptive Golomb_Rice coding for the dyadic sequence was used to entropy code effectively. In addition, the uniform scalar quantization with dead-zone and adjustable parameter was used. The experiments showed that the new algorithm can decrease the coding and decoding time evidently compared with the JPEG2000 algorithm, while maintains favorable loss compression performance.