波段相关性判定和SPIHT的MODIS图像无损压缩

Lossless compression of MODIS multispectral images by combining band correlation with SPIHT algorithm

提出一种结合波段相关性判定和SPIHT编码的方法对MODIS多光谱图像进行无损压缩。首先计算各波段间的相关性,选择相关性强的波段进行预测以去除多光谱图像的谱间冗余;然后对预测后的残差图像进行整数小波变换以去除空间冗余,并用SPIHT算法去除小波系数间的冗余,最后进行算术熵编码。实验结果表明,该方法可对MODIS多光谱图像进行高效无损压缩,优于未进行相关性判定的压缩算法和流行的WinRAR压缩软件。

A lossless compression method based on SPIHT and inter-band correlation prediction was presented for MODIS multispectral images. Firstly, the correlation coefficients among bands were calculated. Secondly, spectral redundancy could be removed by using an optimal linear predictor. Then, spatial redundancy was removed by using the SPIHT algorithm after integer wavelet transform. At last, the arithmetic encoder was used. The achieved compression ratios presented here were compared with those of existing methods. The results show that the proposed lossless compression method for MODIS multispectral images outperforms previous methods.