Circuit and method for determining quantification coefficients in picture compression chains
AssigneeSGS-Thomson Microelectronics S.A.
InventorMartin Bolton
Life of this patent
Estimating the term…
Abstract
A circuit determines a quantification coefficient to be provided to a quantizer incorporated in a picture macroblock compression chain. A first counter counts the number of bits of each macroblock compressed by the compression chain. A first accumulator integrates the difference between each number provided by the first counter and a target, and provides a value corresponding to the quantification coefficient. A microprocessor provides the first accumulator with an initial value and a target at each N macroblocks. Parameters that characterize the current compression process are determined so that the microprocessor determines the target and the initial value to be provided to the first accumulator.
Claims37
A circuit for determining a quantization coefficient to be provided to a quantizer in a picture macroblock compression chain, the circuit comprising:
- a first counter for counting a number of bits of each macroblock compressed by the compression chain;
- a first accumulator, coupled to the first counter, for accumulating from an initial value a difference between each number provided by the first counter and a target value, and for providing as an output a current value corresponding to the quantization coefficient based upon the difference accumulated from initial value.;
- a microprocessor programmed to provide the first accumulator with the initial value, at least at each picture, and a target value at each N macroblocks, where N is a number lower than the number of macroblocks contained in a picture;
- and
- means, coupled to the microprocessor, for determining parameters characterizing the current compression process, the microprocessor being programmed to determine the target value and the initial value provided to the first accumulator from the parameters.
An apparatus for determining a series of quantization coefficients to be used by a data compressor, the data compressor providing a series of bit counts that each represents an amount of compression achieved by the data compressor, the apparatus comprising:
- a bit count difference accumulator, having a first input that receives the series of bit counts, a second input that receives a target value, and an output that provides an accumulated difference between the target value and each bit count;
- a gain circuit, having an input coupled to the output of the bit count difference accumulator and an output that provides the series of quantization coefficients by multiplying the accumulated difference by a gain value;
- and
- means for adjusting the target value and for providing the target value to the bit count accumulator.
An apparatus for determining a series of quantization coefficients to be used by a data compressor, the data compressor providing a series of bit counts that each represents an amount of compression achieved by the data compressor, the apparatus comprising:
- a bit count difference accumulator, having a first input that receives the series of bit counts, a second input that receives a target value, and an output that provides an accumulated difference between the target value and each bit count;
- a gain circuit, having an input coupled to the output of the bit count difference accumulator and an output that provides the series of quantization coefficients by multiplying the accumulated difference by a gain value;
- and
- a target value adjustment circuit having a first input that receives the series of quantization coefficients from the gain circuit and a second input that receives the series of bit counts, the target value adjustment circuitry providing to the bit count difference accumulator the target value, the target value being based on the series of quantization coefficients and the series of bit counts.
A method for determining a series of quantization coefficients to be used by a data compressor, the data compressor providing a series of bit counts that each represents an amount of compression achieved by the data compressor, the method comprising the steps of:
- accumulating a difference between a target value and each bit count;
- multiplying the accumulated difference by a gain value to yield one of the series of quantization coefficients;
- and
- adjusting the target value.
Dependencies are inferred from the claim text.