L. Guo and Y. Meng (USA)
Human visual system, perceptual image coding, wavelet transform, adaptive quantization
An HVS-based wavelet image coder has been proposed in this paper. This coder exploits a couple of important human visual properties, including contrast sensitivity, luminance adaptation, and contrast masking. As the rudiment of the proposed scheme, the proper evaluation of local background luminance and contrast in wavelet domain is investigated within the dyadic wavelet transform framework. Furthermore, the mechanisms of several contrast masking effects, e.g., self-masking, neighbor masking, and cross-band masking are studied and incorporated into the coding scheme through an adaptive quantization scheme. Meanwhile, the coefficient dependent adaptive step sizes of the quantization can be lossless restored during the decoding without any overhead of side information.
Important Links:
Go Back
