A PCA-FVQ and 3D-SPECK combined hierarchical coding method for spectral imagery

Abstract

In this paper, a hierarchical spectral imagery coding method has been proposed. The spectral imagery is encoded into two layers, i.e. base layer and enhancement layer. Firstly, an PCA-FVQ (Principal Component Analysis based Fast Vector Quantization) coding method has been proposed to generate the base layer bitstream, which can be decoded independently to provide a basic image quality. Then the differential data between the original image and the reconstructed image decoded from the base layer bitstream is encoded by 3D-SPECK algorithm to generate the enhancement layer bitstream which can provide an enhanced image quality. The experimental results show that, under the same coding conditions, compared with 3D-SPECK algorithm, the proposed method has achieved 1-15dB higher image quality improvement in terms of SNR (Signal to Noise) while the compression ratio can be controlled precisely.