提出了基于KLT/WT和谱特征矢量量化(SFCVQ)三维谱像数据压缩的新方法。在对多光谱图像数据进行KarhunenLeove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SFCVQ编码对每个谱像数据进行压缩, 获得较高的压缩性能。实验结果表明:KLT/WT/SFCVQ方法和KLT/WT/VQ压缩方法比在同样压缩比(CR)条件下, 峰值信噪比(PSNR)没明显变化, 而速度提高了30倍, 比KLT/WT/FSVQ也提高了5倍, 整体压缩性能有较大的提高。

In this paper we propose a new method for multispec tr al image data compression based on Karhunen-Leove Transformation (KLT)/Wavelet Transformation (WT) and VQ with Spectral Feature Coding(SFCVQ). After KLT is app lied to multispectral image data for exploiting the spectral correlation, the Wa velet Transformation (WT) is used for the transformed multispectral image data t o remove spatial redundancy. Then the SFCVQ is designed to compress every spectr al image data. A higher compression performance is obtained. Experimental result s shows that in comparison of the methods of KLT/WT/SFCVQ with KLT/KLT/VQ, under the condition of the same Compression Ratio (CR), the Peak Signal to Noise Ratio (PSNR) is not varied apparently, while the compression speed increase s 30folds, or 5folds compared with the KLT/WT/FSVQ, and the total compression pe rformance has a great enhancement.