Quantization plays special important roles in image compression system

and its structure and parameters directly influence the quality of reconstructed image and the compression ratio.In our study of Compression System of 3D hyperspectral images

we have disigned two quantization:one as of 1-D arctangnet mapping quantization

the other as of 2-D image edge retaining quantization.The experiment results showed that mapping quantizing error in mapping quantization was low and could better retain the edge feature in hyperspectral image.