Methods and apparatus for improved quantization rounding offset adjustment for video encoding and decoding
AssigneeThomson Licensing
InventorsXiaoan Lu, Qian Xu, Cristina Gomila
Life of this patent
Estimating the term…
Abstract
There are provided methods and apparatus for improved quantization rounding offset adjustment for video encoding and decoding. An apparatus includes a video encoder for encoding an input coding unit by determining a residual for the input coding unit corresponding to a difference between the input coding unit and a reference coding unit, applying a transform to the residual to obtain at least one transform coefficient, and quantizing the at least one transform coefficient by selecting a particular quantization step size and a rounding offset.
Claims21
An apparatus, comprising:
- a video encoder for encoding an input coding unit by determining a residual for the input coding unit corresponding to a difference between the input coding unit and a reference coding unit, applying a transform to the residual to obtain at least one transform coefficient, and quantizing the at least one transform coefficient by selecting a quantization step size and a rounding offset wherein the rounding offset is selected and applied only for smooth classified macroblocks while a default rounding offset typically used by said encoder is applied to textured classified macroblocks.
In a video encoder, a method for encoding an input coding unit comprising:
- determining a residual for the input coding unit corresponding to a difference between the input coding unit and a reference coding unit;
- applying a transform to the residual to obtain at least one transform coefficient;
- and
- quantizing the at least one transform coefficient by selecting a quantization step size and a rounding offset wherein the rounding offset is selected and applied only for smooth classified macroblocks while a default rounding offset typically used by said encoder is applied to textured classified macroblocks.
An apparatus, comprising:
- a video decoder for decoding a coding unit by receiving at least one quantized coefficient corresponding to a transformed and quantized residual between an original version of the coding unit and a reference coding unit, de-quantizing the at least one quantized coefficient to obtain at least one de-quantized coefficient by determining a quantization step size and a rounding offset, and applying an inverse transform to the at least one de-quantized coefficient to obtain a reconstructed residual,
- wherein said video decoder receives information describing the selected rounding offset from a bitstream that includes the at least one quantized coefficient wherein the rounding offset is selected and applied only for smooth classified macroblocks while a default rounding offset typically used by said encoder is applied to textured classified macroblocks.
In a video decoder, a method for decoding a coding unit comprising:
- receiving at least one quantized coefficient corresponding to a transformed and quantized residual between an original version of the coding unit and a reference coding unit;
- de-quantizing the at least one quantized coefficient to obtain at least one de-quantized coefficient by selecting a quantization step size and a rounding offset;
- and
- applying an inverse transform to the at least one de-quantized coefficient to obtain a reconstructed residual,
- wherein said video decoder receives information describing the selected rounding offset from a bitstream that includes the at least one quantized coefficient wherein the rounding offset is selected and applied only for smooth classified macroblocks while a default rounding offset typically used by said encoder is applied to textured classified macroblocks.
A nontransitory computer readable storage media having video signal data encoded thereupon, comprising:
- an input coding unit encoded by determining a residual for the input coding unit corresponding to a difference between the input coding unit and a reference coding unit, applying a transform to the residual to obtain at least one transform coefficient, and quantizing the at least one transform coefficient by selecting a quantization step size and a rounding offset wherein the rounding offset is selected and applied only for smooth classified macroblocks while a default rounding offset typically used by said encoder is applied to textured classified macroblocks.
Dependencies are inferred from the claim text.