Channel scaling for rounding noise reduction in minimum lifting 3D wavelet transform

An integer transform is used in lossless-lossy coding since it can reconstruct an input signal without any loss at output of the backward transform. Recently, its number of lifting steps is reduced as well as delay from input to output introducing multi-dimensional memory accessing. However it has a problem that quality of the reconstructed signal in lossy coding has its upper bound in the rate distortion curve. This is because the noise generated by rounding operations in each lifting step inside the integer transform does not contribute to data compression. This paper tries to reduce the rounding noise observed at output of the integer transform introducing channel scaling inside the transform. As a result of experiments, it was observed that the proposed method improves quality of the decoded signal in lossy coding mode.