Optimized subband coding of images

Abstract

This paper presents recent results in analysis and design of optimum two dimensional (2D) filter banks for subband coding of images in the presence of quantization noise. General M-band 2D separable filter banks are considered. The quantization effects in two dimensional subband coding system is analyzed by employing the polyphase decomposition technique of the subband filter bank and using the gain-plus-additive noise model for the quantizer. The mean square (MS) quantization error due to the quantizer if then calculated at the system output. This mean square error (MSE) serves as the measure in the optimum design and evaluation of subband filter banks. Numerical design examples with orthonormal and biorthonormal constraints on the subband filter banks are shown and tested with actual LENA image.