Image coding using visual primary component and wavelet transform

In this paper a visual primary component (VPC) model is proposed. By applying the LoG operator, VPC can be extracted and coded effectively according to the edge information such as strength, length, integrity and contrast with the background. The residual image is then compressed using a biorthogonal wavelet transform. Reconstructed images show that the "ringing effect" existing in single-phase wavelet coders has been greatly reduced. Compared with JPEG, the proposed hybrid scheme performs significantly better both in terms of PSNR and visual quality.