Back to results
US 6,289,128 B1Granted
Method and apparatus for sharing a least significant bit of encoded data between multiple numeric data values
AssigneeRicoh Company, Ltd.
InventorJames D. Allen
Life of this patent
Estimating the term…
Abstract
An encoding technique in which numeric data in a set of data is selected and the least significant bits (LSBs) of numeric data values in the numeric data are replaced with a single least significant bit that is shared by the numeric data values in the numeric data.
Claims79
A method of encoding data, said method comprising:
- selecting numeric data to be quantized;
- and
- replacing the least significant bits (LSBS) of the numeric data with a single LSB shared by each numeric data value in the numeric data, wherein replacing the LSBs comprises providing the single LSB based on mean squared error (MSE).
A method of encoding data, said method comprising:
- selecting numeric data to be quantized;
- and
- replacing the least significant bits (LSBS) of the numeric data with a single LSB shared by each numeric data value in the numeric data, wherein replacing LSBs comprises setting the single LSB according to which binary value occurs most often among the LSBs of the numeric data values in the numeric data.
A method of encoding data, said method comprising:
- selecting numeric data to be quantized;
- and
- replacing the least significant bits (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data, wherein replacing the LSBs comprises setting the single LSB according to which binary value occurs most often among the LSBs over a region of image data containing the numeric data.
A method of encoding data, said method comprising:
- selecting a group of numeric data to be quantized;
- and
- replacing the least significant bits (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data, wherein replacing the LSBs comprises setting the single LSB to a predetermined binary value if ∑ x ∈ group ( 2 x - 2 ⌊ x ⌋ - 1 ) ( 2 ⌊ x ⌋ - 4 ⌊ x 2 ⌋ - 1 ) < 0.
A method of encoding data, said method comprising:
- selecting numeric data to be quantized;
- replacing the least significant bits (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data;
- and
- embedding the single LSB in a file compatible with an unwitting standard, wherein embedding the single LSB in the file comprises attaching the single LSB to one of said each numeric data value.
1 more element
A method of encoding data, said method comprising:
- selecting numeric data to be quantized;
- replacing the least significant bits (LSBS) of the numeric data with a single LSB shared by each numeric data value in the numeric data, and
- embedding the single LSB in a file compatible with an unwitting standard, wherein embedding the single LSB in the file comprises modifying a q-matrix.
A method of encoding data, said method comprising:
- selecting a group of numeric data to be quantized;
- and
- replacing the least significant bits (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data, wherein replacing the least significant bit (LSBs) of the numeric data with a single LSB comprises setting the single LSB to a redetermined binary value if ∑ x ∈ group xC ( [LeftBracketingBar] x [RightBracketingBar] ) [LeftBracketingBar] x [RightBracketingBar] > 0
- where C ( z ) = { 4 zB if z < D 2 z ( B - .75 ) + 9 / 16 - B 2 + K if D < z < D + .5 z - K - ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ = ⌊ z + .5 ⌋ ≥ 1 - z + K + 0.5 + ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ > ⌊ z + .5 - D ⌋ ≥ 1
- and variables B, D and K refer to the reconstruction level for any zero, the bias for quantization thresholds and the appropriately scaled ratio between rate and distortion costs, respectively.
2 more elements
A method of encoding data, said method comprising:
- selecting numeric data to be quantized;
- and
- replacing the least significant bits (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data, wherein replacing the LSBs comprises converting the sign of said each numeric data value by exclusive-ORing the sign with the single LSB.
A method of encoding data, said method comprising:
- selecting numeric data to be quantized;
- and
- replacing the least significant bits (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data, wherein replacing the LSBs comprises converting the sign of each numeric data value by exclusive-ORing the sign with the probability that the sign is in a predetermined state.
An encoder comprising:
- means for receiving numeric data to be quantized;
- and
- means for replacing the least significant bit (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data, wherein the means for replacing comprises means for selecting the single LSB based on mean squared error (MSE).
An encoder comprising:
- means for receiving numeric data to be quantized;
- and
- means for replacing the least significant bit (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data, wherein the means for replacing comprises means for setting the single LSB according to which binary value occurs most often among the LSBs of the numeric data values in the numeric data.
An encoder comprising:
- means for receiving numeric data to be quantized;
- and
- means for replacing the least significant bit (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data, wherein the means for replacing comprises means for setting the single LSB according to which binary value occurs most often among the LSBs over a region of image data containing the numeric data.
An encoder comprising:
- means for receiving a group of numeric data to be quantized;
- and
- means for replacing the least significant bit (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data, wherein the means for replacing comprises means for setting the single LSB to a predetermined binary value if ∑ x ∈ group ( 2 x - 2 ⌊ x ⌋ - 1 ) ( 2 ⌊ x ⌋ - 4 ⌊ x 2 ⌋ - 1 ) < 0.
An encoder comprising:
- means for receiving numeric data to be quantized;
- means for replacing the least significant bit (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data;
- and
- means for embedding the single LSB in a file compatible with an unwitting standard, wherein the means for embedding the single LSB in the file comprises means for attaching the single LSB to one of said each numeric data value.
1 more element
An encoder comprising:
- means for receiving numeric data to be quantized;
- means for replacing the least significant bit (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data;
- and
- means for embedding the single LSB in a file compatible with an unwitting standard, wherein the means for embedding the single LSB in the file comprises means for modifying a q-matrix.
1 more element
An encoder comprising:
- means for receiving a group of numeric data to be quantized;
- and
- means for replacing the least significant bit (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data, wherein the means for replacing the least significant bit (LSBs) of the numeric data with a single LSB comprises means for setting the single LSB to a predetermined binary value if ∑ x ∈ group xC ( [LeftBracketingBar] x [RightBracketingBar] ) [LeftBracketingBar] x [RightBracketingBar] > 0
- where C ( z ) = { 4 zB if z < D 2 z ( B - .75 ) + 9 / 16 - B 2 + K if D < z < D + .5 z - K - ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ = ⌊ z + .5 ⌋ ≥ 1 - z + K + 0.5 + ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ > ⌊ z + .5 - D ⌋ ≥ 1
- and variables B, D and K refer to the reconstruction level for any zero, the bias for quantization thresholds and the appropriately scaled ratio between rate and distortion costs, respectively.
2 more elements
An encoder comprising:
- means for receiving numeric data to be quantized;
- and
- means for replacing the least significant bit (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data;
- wherein the means for replacing the LSBs comprises means for converting the sign of said each numeric data value by exclusive-ORing the sign with the single LSB.
1 more element
An encoder comprising:
- means for receiving numeric data to be quantized;
- and
- means for replacing the least significant bit (LSBs) of the numeric data with a single LSB shared by each numeric data value in the numeric data, wherein the means for replacing the LSBs comprises means for converting the sign of each numeric data value by exclusive-ORing the sign with the probability that the sign is in a predetermined state.
An encoder comprising:
- a memory to store a plurality of numeric data values;
- and
- a least significant bit (LSB) processing logic coupled to the memory to replace LSBs of the plurality of numeric data values with a single LSB shared by the numeric data values, wherein the LSB processing logic provides the single LSB based on mean squared error (MSE).
An encoder comprising:
- a memory to store a plurality of numeric data values;
- and
- a least significant bit (LSB) processing logic coupled to the memory to replace LSBs of the plurality of numeric data values with a single LSB shared by the numeric data values, wherein the LSB processing logic sets the single LSB according to which binary value occurs most in the LSBs of the plurality of numeric data values.
An encoder comprising:
- a memory to store a group of numeric data values;
- and
- a least significant bit (LSB) processing logic coupled to the memory to replace LSBs of the group of numeric data values with a single LSB shared by the numeric data values, wherein the LSB processing logic sets the single LSB to a predetermined binary value if ∑ x ∈ group ( 2 x - 2 ⌊ x ⌋ - 1 ) ( 2 ⌊ x ⌋ - 4 ⌊ x 2 ⌋ - 1 ) < 0.
An encoder comprising:
- a memory to store a group of numeric data values;
- and
- a least significant bit (LSB) processing logic coupled to the memory to replace LSBs of the group of numeric data values with a single LSB shared by the numeric data values, wherein the LSB processing logic sets the single LSB to a predetermined binary value if ∑ x ∈ group xC ( [LeftBracketingBar] x [RightBracketingBar] ) [LeftBracketingBar] x [RightBracketingBar] > 0
- where C ( z ) = { 4 zB if z < D 2 z ( B - .75 ) + 9 / 16 - B 2 + K if D < z < D + .5 z - K - ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ = ⌊ z + .5 - D ⌋ ≥ 1 - z + K + 0.5 + ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ > ⌊ z + .5 - D ⌋ ≥ 1
- and variables B, D and K refer to the reconstruction level for any zero, the bias for quantization thresholds and the appropriately scaled ratio between rate and distortion costs, respectively.
2 more elements
An encoder comprising:
- a memory to store a plurality of numeric data values;
- and
- a least significant bit (LSB) processing logic coupled to the memory to replace LSBs of the plurality of numeric data values with a single LSB shared by the numeric data values, the LSB processing logic to convert the sign of said each numeric data value by exclusive-ORing the sign with the single LSB.
An encoder comprising:
- a memory to store a plurality of numeric data values;
- and
- a least significant bit (LSB) processing logic coupled to the memory to replace LSBs of the plurality of numeric data values with a single LSB shared by the numeric data values, the LSB processing logic to convert the sign of each numeric data value by exclusive-ORing the sign with the probability that the sign is in a predetermined state.
A computer product comprising a sequence of instructions stored on a computer readable medium which, when executed by at least one processing device in a system, cause said at least one processing device to:
- select numeric data in a set of data;
- and
- replace the least significant bit (LSBs) of the numeric data with a single LSB to be shared by numeric data values in the numeric data by providing the single LSB based on mean squared error (MSE).
A computer product comprising a sequence of instructions stored on a computer readable medium which, when executed by at least one processing device in a system, cause said at least one processing device to:
- select numeric data in a set of data;
- and
- replace the least significant bit (LSBs) of the numeric data with a single LSB to be shared by numeric data values in the numeric data by setting the single LSB according to which binary value occurs most often among the LSBs of the numeric data values in the numeric data.
A computer product comprising a sequence of instructions stored on a computer readable medium which, when executed by at least one processing device in a system, cause said at least one processing device to:
- select numeric data in a set of data:
- and
- replace the least significant bit (LSBs) of the numeric data with a single LSB to be shared by numeric data values in the numeric data by setting the single LSB according to which binary value occurs most often among the LSBs over a region of image data containing the numeric data.
A computer product comprising a sequence of instructions stored on a computer readable medium which, when executed by at least one processing device in a system, cause said at least one processing device to:
- select numeric data in a group of data;
- and
- replace the least significant bit (LSBs) of the numeric data with a single LSB to be shared by numeric data values in the numeric data by setting the single LSB to a predetermined binary value if ∑ x ∈ group ( 2 x - 2 ⌊ x ⌋ - 1 ) ( 2 ⌊ x ⌋ - 4 ⌊ x 2 ⌋ - 1 ) < 0.
A computer product comprising a sequence of instructions stored on a computer readable medium which, when executed by at least one processing device in a system, cause said at least one processing device to:
- select numeric data in a set of data;
- and
- replace the least significant bit (LSBs) of the numeric data with a single LSB to be shared by numeric data values in the numeric data by converting the sign of said each numeric data value by exclusive-ORing the sign with the single LSB.
A computer product comprising a sequence of instructions stored on a computer readable medium which, when executed by at least one processing device in a system, cause said at least one processing device to:
- select numeric data in a set of data;
- and
- replace the least significant bit (LSBs) of the numeric data with a single LSB to be shared by numeric data values in the numeric data by converting the sign of each numeric data value by exclusive-ORing the sign with the probability that the sign is in a predetermined state.
A computer product comprising a sequence of instructions stored on a computer readable medium which, when executed by at least one processing device in a system, cause said at least one processing device to:
- select numeric data in a group of data;
- and
- replace the least significant bit (LSBs) of the numeric data with a single LSB to be shared by numeric data values in the numeric data by setting the single LSB to a predetermined binary value if ∑ x ∈ group xC ( [LeftBracketingBar] x [RightBracketingBar] ) [LeftBracketingBar] x [RightBracketingBar] > 0
- where C ( z ) = { 4 zB if z < D 2 z ( B - .75 ) + 9 / 16 - B 2 + K if D < z < D + .5 z - K - ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ = ⌊ z + .5 - D ⌋ ≥ 1 - z + K + 0.5 + ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ > ⌊ z + .5 - D ⌋ ≥ 1
- and variables B, D and K refer to the reconstruction level for any zero, the bias for quantization thresholds and the appropriately scaled ratio between rate and distortion costs, respectively.
2 more elements
A decoder comprising:
- a receiver to receive a group of quantized numeric data sharing a single LSB;
- and
- a decoder coupled to the receiver to decode the quantized numeric data using the single LSB for each numeric data values, wherein the decoder decodes data that has been encoded by replacing the least significant bit (LSBs) of the numeric data with a single shared by each numeric data values in the group of numeric data by setting the single LSB to a predetermined binary value if ∑ x ∈ group xC ( [LeftBracketingBar] x [RightBracketingBar] ) [LeftBracketingBar] x [RightBracketingBar] > 0
- where C ( z ) = { 4 zB if z < D 2 z ( B - .75 ) + 9 / 16 - B 2 + K if D < z < D + .5 z - K - ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ = ⌊ z + .5 - D ⌋ ≥ 1 - z + K + 0.5 + ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ > ⌊ z + .5 - D ⌋ ≥ 1
- and variables B, D and K refer to the reconstruction level for any zero, the bias for quantization thresholds and the appropriately scaled ratio between rate and distortion costs, respectively.
2 more elements
A method of decoding data in which a plurality of encoded data values share a single LSB comprising:
- generating a reconstructed data value for each of the plurality of encoded data values using the single LSB;
- wherein generating the reconstructed data is performed according to the following equation:
- {overscore (x)}y=2yi+L−½
- where xi is the reconstructed data value, yi is the most significant bit of the encoded data value being reconstructed and L is the single LSB being shared;
- and
- emitting the reconstructed data for each of the plurality of encoded data.
3 more elements
A method of encoding comprising:
- selecting a group of numeric data to be quantized;
- and
- replacing the least significant bit (LSBs) of the numeric data with a single shared LSB by each numeric data values in the group of numeric data, wherein replacing the LSBs comprises setting the single LSB to a predetermined binary value L if ∑ x ∈ group xC ( [LeftBracketingBar] x [RightBracketingBar] ) [LeftBracketingBar] x [RightBracketingBar] > 0
- where C ( z ) = { 4 zB if z < D 2 z ( B - .75 ) + 9 / 16 - B 2 + K if D < z < D + .5 z - K - ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ = ⌊ z + .5 - D ⌋ ≥ 1 - z + K + 0.5 ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ > ⌊ z + .5 - D ⌋ ≥ 1
- and variables B, D and K refer to the reconstruction level for any zero, the bias for quantization thresholds and the appropriately scaled ratio between rate and distortion costs, respectively.
2 more elements
An encoder comprising:
- means for selecting a group of numeric data to be quantized;
- and
- means for replacing the least significant bit (LSBs) of the numeric data with a single shared LSB by each numeric data values in the group of numeric data, wherein replacing the LSBs comprises setting the single LSB to a predetermined binary value L if ∑ x ∈ group xC ( [LeftBracketingBar] x [RightBracketingBar] ) [LeftBracketingBar] x [RightBracketingBar] > 0
- where C ( z ) = { 4 zB if z < D 2 z ( B - .75 ) + 9 / 16 - B 2 + K if D < z < D + .5 z - K - ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ = ⌊ z + .5 - D ⌋ ≥ 1 - z + K + 0.5 ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ > ⌊ z + .5 - D ⌋ ≥ 1
- and variables B, D and K refer to the reconstruction level for any zero, the bias for quantization thresholds and the appropriately scaled ratio between rate and distortion costs, respectively.
2 more elements
A computer product comprising a sequence of instructions stored on a computer readable medium which, when executed by at least one processing device in a system, cause said at least one processing device to:
- select a group of numeric data to be quantized;
- and
- replace the least significant bit (LSBs) of the numeric data with a single shared LSB by each numeric data values in the group of numeric data, wherein replacing the LSBs comprises setting the single LSB to a predetermined binary value L if ∑ x ∈ group xC ( [LeftBracketingBar] x [RightBracketingBar] ) [LeftBracketingBar] x [RightBracketingBar] > 0
- where C ( z ) = { 4 zB if z < D 2 z ( B - .75 ) + 9 / 16 - B 2 + K if D < z < D + .5 z - K - ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ = ⌊ z + .5 ⌋ ≥ 1 - z + K + 0.5 + ⌊ z + .5 - D ⌋ if ⌊ z + 1 - D ⌋ > ⌊ z + .5 - D ⌋ ≥ 1
- and variables B, D and K refer to the reconstruction level for any zero, the bias for quantization thresholds and the appropriately scaled ratio between rate and distortion costs, respectively.
2 more elements
Dependencies are inferred from the claim text.