结合预测误差反馈的高光谱图像无损压缩研究

Lossless Compression for Hyperspectral Images Combined with Predictive Error Feedback

高光谱图像的海量数据给存储和实时传输带来极大困难,必须对其进行有效压缩。提出了一种结合预测误差反馈的高光谱图像无损压缩算法。根据高光谱图像相邻波段相关性强弱进行波段分组,有效降低了波段排序算法的计算量。通过研究波段排序算法的性能,采用最佳后向排序算法对各组进行波段排序。为有效去除高光谱图像相关性,采用JPEG压缩标准中的无损预测模式对各波段进行谱内预测,利用参考波段预测误差对当前波段谱内预测值进行反馈校正,可进一步提高预测精度。最后,利用JPEG-LS标准对参考波段和预测残差进行无损压缩。对AVIRIS型和OMIS-Ⅰ型高光谱图像的实验结果表明,该算法可显著降低压缩后的平均比特率。

The data size of hyperspectral images is too large for storage or transmission, so it is necessary to compress hyperspec- tral images efficiently. A new lossless compression algorithm for hyperspectral images combined with predictive error feedback is pro- posed. To decrease the reordering computational complexity, spectral band grouping algorithm is introduced to divide hyperspectral ima- ges into several groups according to the correlation coefficient between each adjacent bands, then each group is reordered by using the best reverse reordering algorithm based on the performance analysis of several reordering algorithms. To remove the redundancy efficient- ly, JPEG lossless compression modes are used for intra-prediction of each band, while the predictive errors of reference band are used to revise the intra-prediction values of current band,the final predictive errors are compressed by JPEG-LS standard. Experimental results show that the proposed algorithm can give better lossless coding performance.