Visual Assessment for The Quantization Error in Wavelet Based Monochrome Videos

The investigation of the discrete wavelet transform (DWT) based video coder is still undergoing in the literature. One of the open problems to be solved is the perception to the quantization noise in different subbands in the DWT domain. This is a critical issue for the development of a better motion compensation (MC) scheme. An experiment and relevant results analysis are presented in this paper to address the above issue. Monochrome video sequences of natural scenes are used in the experiment therefore the so-called masking effects can be taken into account in the decision of the sensitivity to the noise hidden in the DWT domain. The preliminary results show that the most sensitive subbands are those in the lowest three resolution levels under a five-levels decomposition scheme. The further analysis proves that the distribution of the sensitivity to each individual subband has been shifted by the context of the video.