基于KLT／WT／WTVQ的三维多光谱数据压缩方法

提出了三维多谱图像数据压缩的新方法。在对多光谱图像数据进行Karhunen-Leove变换（KLT）消除谱相关性，再应用小波变换（WT）对KLT后的多光谱图像数据进行消除空间相关性。采用特殊矢量量化编码，叫作小波树矢量量化（WTVQ）编码对每个谱象数据进行压缩，获得较高的压缩性能。实验结果表明：KLT／WT／WTVQ压缩方法比KLT／JPEG和KLT／WT／FSVQ压缩方法在相同的PSNR条件下，CR提高20以上，整体压缩性能有较大的提高；在同样压缩比（CR）条件下，压缩速度提高3倍。

A Method for 3D Multispectral Data Compression Based on KLT/WT/WTVQ

In this paper, a new method for multispectral image data was proposed. After Karhunen leove transformation (KLT) was applied in multispectral image d ata to exploit the spectral correlation, Wavelet Transformation(WT) was used fo r the transformed multispectral image data to remove spatial redundancy. Then a special Vector Quantization, called Wavelet Tree Vector Quantization(WTVQ), is designed to compress the every spectral image data, and a higher compression per formance is obtained. Experiment results shows that, in comparison of thi s method of KLT/WT/WTVQ with the method of KLT/JPEG and KLT/WT/FSVQ, the CR is i ncreased 20folds in the case of same PSNR, and the total compression performance has a great enhancement.;the compression speed is enhanced tree times in the c ase of same Compression Ratio(CR).