论文部分内容阅读
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。
A new method based on KLT / WT and spectral feature vector quantization (SFCVQ) is proposed. The Karhunen-Leove transform (KLT) is used to eliminate the spectral correlation of multi-spectral image data, and the wavelet transform (WT) is used to eliminate the spatial correlation of multi-spectral image data after KLT. Using SFCVQ coding to compress each spectral data to obtain higher compression performance. The experimental results show that the PSNR of KLT / WT / SFCVQ method and KLT / WT / VQ method do not change obviously under the same compression ratio (CR) WT / FSVQ also increased 5 times, the overall compression performance has been greatly improved.