Quantizing apparatus and quantizing method

A quantizing apparatus for quantizing a difference signal representing the difference between a sample value of an input digital signal and a predicted value, comprising:

• means for obtaining a difference value between one of representative values of which the difference signal is quantized and a most concentrated value of a signal value to be quantized;
• means for shifting a quantizing representative value corresponding to the difference value so that the signal to be quantized accords with a most concentrated value of the signal to be quantized;
• and
• means for performing a quantizing operation corresponding to the shifted representative value,
• wherein an output signal of said quantizing means is transmitted, and
• wherein both a code corresponding to a representative value of which the difference signal is quantized and data defining a quantizing condition are transmitted as side information so that a restored value of the reference value is not changed.

A quantizing method for quantizing a difference signal representing the difference between a sample value of an input digital signal and a predicted value, wherein said difference signal has most concentrated distribution for a portion of said entire input digital signal different from said distribution for the entire input digital signal, said method comprising the steps of:

• obtaining a difference value between one of representative values of which the difference signal is quantized and said most concentrated distribution to be quantized:
• shifting a quantizing representative value corresponding to the difference value so that the signal to be quantized accords with said most concentrated distribution of the signal to be quantized;
• and
• performing a quantizing operation corresponding to the shifted representative value.

A quantizer, comprising:

• level detecting means for forming a block composed of a plurality of spatially and/or temporally adjacent pixels and for detecting a particular level of which the pixels of the block are concentrated;
• compensation amount generating means for generating a compensation amount corresponding to the detected particular level;
• compensating means for compensating a reference value of the block with the compensation amount;
• normalizing means for normalizing the pixels of the block corresponding to the reference value;
• and
• means for quantizing the normalized pixels.

A method for quantizing pixel information, comprising the steps of:

• forming a block composed of a plurality of spatially and/or temporally adjacent pixels and detecting a particular level of which the pixels of the block are concentrated;
• generating a compensation amount corresponding to the detected particular level;
• compensating a reference value of the block with the compensation amount;
• normalizing the pixels of the block corresponding to the reference value;
• and
• quantizing the normalized pixels.