基于自适应预测的高光谱遥感图像无损压缩算法

针对高光谱遥感图像细节丰富纹理复杂、空间相关性弱、难于压缩的特点，充分利用高光谱遥感图像的谱间相关性,使用多个波段的像素来自适应预测当前波段的像素。因为待预测像素的最优预测是其条件期望，用分段积分的方法将条件期望转化为可计算的表达式，并与其它波段的像素关联起来。选取与待预测像素有较强因果关系的相邻像素自适应地估计出各参数的值，得到残差图像，消除了大部分的谱间冗余和空间冗余，再用JPEG-LS进一步去除残差图像的空间相关性。实验表明，该算法能有效去除高光谱图像间的相关性，较其它压缩算法压缩比有很大提高，且算法简单，便于硬件实现。


Lossless Compression of Hyperspectral Image Based on Adaptive Prediction

Hyperspectral images are hard to compress because of their abundant details,complicated texture and insignificant special correlation.Making use of the significant spectral correlation within the hyperspectral images,We use pixels of several bands to adaptively predict the pixels of the current band.We can know the optimal prediction of the current pixels is its conditional expectation,which can be translated into computational expression by using subsection ingtegral,and associate with pixels of other bands.We choose neighboring pixels to adaptive estimate every parameter in order to get residual image,which remove most spatial and spectral redundancy,and then we use JPEG-LS to remove the spectral redundancy.Experiments show that the algorithm can compress the data efficiently and works better than other algorithms.