高光谱图像的分布式近无损压缩

Distributed Near Lossless Compression of Hyperspectral Images

星载高光谱图像的有效压缩已经成为高光谱遥感领域亟待解决的难题。分布式信源编码具有较低的编码复杂度与良好的抗误码性,在高光谱图像压缩领域具有广阔的应用前景。提出了一种基于多元陪集码的高光谱图像分布式近无损压缩算法。根据多元陪集码的Slepian-Wolf无损编码的压缩过程,提出了面向高光谱图像分布式近无损压缩的最优量化方案,使得高光谱图像在给定目标码率条件下的失真达到最小,在此基础上对量化值进行Slepian-Wolf无损编码,从而实现了高光谱图像的分布式近无损压缩。实验结果表明,与典型的传统算法相比,该算法取得了较好的近无损压缩性能和较低的编码复杂度。

The efficient compression of onboard hyperspectral images has been a difficult problem which needs to be resolved urgently. Low encoding complexity and excellent error resilience are provided by distributed source coding, which has wide applied foreground in the field of hyperspectral images compression. For the problem of onboard compression for hyperspectral images, a distributed near lossless compression algorithm based on multi-level coset codes is proposed. According to the procedure of SlepianWolf lossless coding based on multi- level coset codes, an optimal quantization scheme for distributed near lossless compression of hyperspectral images is presented, which makes the distortion of hyperspectral images minimum under the given target bit-rates. Slepian-Wolf lossless coding is performed on the quantized values,which realizes the distributed near lossless compression of hyperspectral images. Experimental results show that the proposed algorithm can obtain both high near lossless compression performance and low encoding complexity compared with those existed classical algorithms.