METHOD AND DEVICE FOR MINIMISING A QUANTISATION ERROR

摘要

A device minimizes a quantization error in the inverse quantization of a quantized coefficient during a compression method, said quantization error describing a difference between the quantized coefficient after inverse quantization and an associated coefficient prior to quantization. According to said method, at least one parameter that is characteristic of the compression of the quantized coefficient is selected, (at least two characteristic parameters being selected if one of the two characteristic parameters corresponds to a temporal prediction mode), a correction value is chosen based on the selected characteristic parameter(s) and inverse quantization takes place after the correction value has been added to the quantized coefficient.