一种适合实时应用的多光谱遥感图像无损压缩算法

A Lossless Compression Arithmetic of Multi-spectral Image Data

根据多（超）光谱成像光谱仪波段数众多，不同波段间图像光谱相关性较强的特点，在使用预测编码（ＤＰＣＭ）方法去相关时，先针对某波段生成自适应最佳预测模型，并将此最佳预测模型应用到所有波段，然后再进行光谱维ＤＰＣＭ进一步去除光谱间相关性，最后使用预测误差修正技术来减小预测残差冗余度．实验表明，这种无损压缩方法效果大大优于ＪＰＥＧ无损压缩标准方法，并且速度很快，实现简便，适用于实时系统．

Due to the high spectral correlation of multi-spectral image, author use an inter-frame auto adjusting predict model of a special band image in all bands to get rid of spatial correlation, and a spectral reflection feature based error correction model to get rid of spectral correlation. The effect of this compression arithmetic is better than JPEG (lossless) and has low complexity and low requirement of hardware processing ability