Back to results
US 8,135,226 B2Granted
Image encoder, image encoding method, image decoder, and image decoding method
AssigneeSony Corporation
InventorsTakahiro Fukuhara, Seiji Kimura
Life of this patent
Estimating the term…
Abstract
In an image encoder, a first quantization section which is selected when normal image quality is required performs quantization by dividing a wavelet transformation coefficient by a quantization step size and by thereafter rounding down a fraction thereof. On the other side, a second quantization section which is selected when high image quality is required performs quantization by dividing a wavelet transformation coefficient by a quantization step size, by adding 0.5 to the addition result, and by thereafter rounding down a fraction thereof. Therefore, the width of a dead zone where coefficients are quantized to a value of 0 is narrower than that in the first quantization section, and higher image quality is obtained accordingly.
Claims8
An image encoder comprising:
- filtering means which hierarchically performs a filtering process on an input image, to generate a plurality of sub-bands;
- quantization means which quantizes coefficients in each sub-band after the filtering processing, wherein each sub-band has an assigned quantization step size;
- code block generation means which divides the sub-bands, to generate a plurality of code blocks each having a predetermined size;
- bit plane generation means which generates bit planes from a most significant bit to a least significant bit, for each code block;
- encoding means which arithmetically encodes a coding pass generated for each bit plane;
- and
- format means which formats by adding a header to an arithmetic code generated by the encoding means, to generate an encoded code stream,
- wherein,
- the quantization means has a capability to vary a width of a dead zone corresponding to the assigned quantization step size by (1) dividing each of the coefficients in each of the sub-bands by the assigned quantization step size corresponding to the sub-band to which the coefficient being divided belongs and (2) adding a first constant value to the result of the division before rounding down the result, the dead zone being a quantization zone where the coefficients are quantized to a value of 0, and
- the format means sets a predetermined parameter included in the header with the quantization step size, the predetermined parameter being used to determine a second constant value for use with the quantization step size to calculate a wavelet transformation coefficient when an inverse quantization corresponding to the quantization of the quantization means is performed;
- and
- wherein
- when the width of a dead zone has been varied, then the second constant value is 0;
- otherwise the second constant value is determined as one of a plurality of values between and including zero and one, and said plurality of values include values other than 0 and 0.5.
11 more elements
An image encoding method comprising:
- hierarchically performing a filtering process on an input image, to generate a plurality of sub-bands;
- quantizing coefficients in each sub-band after the filtering processing, wherein each sub-band has an assigned quantization step size;
- dividing the sub-bands, to generate a plurality of code blocks each having a predetermined size;
- generating bit planes from a most significant bit to a least significant bit, for each code block;
- arithmetically encoding a coding pass generated for each bit plane;
- and
- performs formatting by adding a header to an arithmetic code generated by the encoding step, to generate an encoded code stream,
- wherein,
- the step of quantizing optionally varies a width of a dead zone corresponding to the assigned quantization step size by (1) dividing each of the coefficients in each of the sub-bands by the assigned quantization step size corresponding to the sub-band to which the coefficient being divided belongs and (2) adding a first constant value to the result of the division before rounding down the result, the dead zone being a quantization zone where the coefficients are quantized to a value of 0, and
- in the formatting step a predetermined parameter is included in the header with the quantization step size, the predetermined parameter being used to determine a second constant value for use with the quantization step size to calculate a wavelet transformation coefficient when an inverse quantization corresponding to the quantization of the quantization means is performed;
- and
- wherein
- when the width of a dead zone has been varied, then the second constant value is 0;
- otherwise the second constant value is determined as one of a plurality of values between and including zero and one, and said plurality of values include values other than 0 and 0.5.
11 more elements
An image encoder comprising:
- a filter for hierarchically performing a filtering process on an input image, to generate a plurality of sub-bands;
- a quantization section for quantizing coefficients in each sub-band after the filtering processing, wherein each sub-band has an assigned quantization step size;
- a code block generator for dividing the sub-bands, to generate a plurality of code blocks each having a predetermined size;
- a bit plane generator for generating bit planes from a most significant bit to a least significant bit, for each code block;
- an encoder for arithmetically encoding a coding pass generated for each bit plane;
- and
- a format section for adding a header to an arithmetic code generated by the encoder, to generate an encoded code stream,
- wherein,
- the quantization section has a capability to vary a width of a dead zone corresponding to the assigned quantization step size by (1) dividing each of the coefficients in each of the sub-bands by the assigned quantization step size corresponding to the sub-band to which the coefficient being divided belongs and (2) adding a first constant value to the result of the division before rounding down the result, the dead zone being a quantization zone where the coefficients are quantized to a value of 0, and
- the format section sets a predetermined parameter included in the header with the quantization step size, the predetermined parameter being used to determine a second constant value for use with the quantization step size to calculate a wavelet transformation coefficient when an inverse quantization corresponding to the quantization of the quantization means is performed;
- and
- wherein
- when the width of a dead zone has been varied, then the second constant value is 0;
- otherwise the second constant value is determined as one of a plurality of values between and including zero and one, and said plurality of values include values other than 0 and 0.5.
11 more elements
Dependencies are inferred from the claim text.