Since the final observer of the processed image is the human, the incorporation of the human visual characteristics into the quantization scheme is more logical than the conventional one which has goal to improve the objective image quality by using the statistical properties.
In this thesis, a weighted quantization scheme for intraframe and interframe difference image signals are presented and proposed, respectively. And several adaptive quantization strategies are also proposed to compensate the quantization artifacts by considering human visual masking effects. It is found that the proposed scheme gives better subjective quality than conventional one.
