An Improved 3D SPIHT Coding for Hyperspectral Imagery Data Compression

In this paper, an improved 3D SPIHT compression coding method, naming 3D BCSPIHT, based on wavelet zero tree compression coding (EZW), is used to exploit redundancy in 3D hyperspectral imagery data. The experimental results show that this method can obtain the compression ratio 200 for 224 spectral band data without any K L transformation (KLT) to remove the spectral correlation, and the total compression performance of this method is better than the method of KLT and other two dimensional transformation (such as 2D DCT and 2DWT) to remove the spectral correlation and spatial redundancy. To compare with the total compression ratio of KLT+JPEG method and KLT+BBZTC method, this method reaches 200 folds at the average PSNR of 32.531 dB. This method enhances 2～5 dB in PSNR at same compression ratio to KLT+BBZTC method, and the average compression rate improves above 15 fold.