Disabling intra prediction filtering

Patent metadata

Field	Value
Publication number	US-9451254-B2
Application number	US-201414334496-A
Country	US
Kind code	B2
Filing date	Jul 17, 2014
Priority date	Jul 19, 2013
Publication date	Sep 20, 2016
Grant date	Sep 20, 2016

Abstract

A video decoder may adaptively disable, based on a syntax element, one or more filters used for intra-prediction. In addition, the video decoder may perform intra-prediction to generate prediction data for a current block of a current video slice. Furthermore, a video encoder may adaptively disable one or more filters used for intra-prediction. Furthermore, the video encoder may signal a syntax element that controls the one or more filters. In addition, the video encoder may perform intra prediction to generate prediction data for a current video block the video data.

First claim

What is claimed is:

1. A method of decoding video data, the method comprising:
adaptively disabling, based on a syntax element specifying whether residual signals are quantized without transformation, one or more filters used for intra-prediction; and
performing the intra-prediction to generate a predictive block for a current block of a current video slice of a current picture of the video data based on a signaled intra prediction mode and data from previously decoded blocks of the current picture, wherein:
the one or more filters include at least one of a mode dependent intra smoothing (MDIS) filter, a bilinear interpolation filter, an edge filter, or a DC intra-prediction mode filter,
the MDIS filter is applied to reference samples differently for different block sizes and intra prediction modes,
when angular prediction is used to generate the predictive block, each respective predicted sample of the predictive block is obtained by projecting a location of the respective sample to a sub-pixel location between two of the reference samples nearest to the sub-pixel location and interpolating the respective predicted sample by applying the bilinear interpolation filter to interpolate at 1/32pixel accuracy using the two reference samples nearest to the sub-pixel location,
the edge filter filters a first prediction column of the predictive block and a first prediction row of the predictive block using gradient based smoothing, and
the DC intra-prediction mode filter filters the first prediction row of the predictive block and the first prediction column of the predictive block using a DC value with a two-tap Finite Impulse Response (FIR) filter and filters a corner sample of the predictive block using a 3-tap filter.

2. The method of claim 1 , wherein the syntax element is signaled in a prediction unit (PU).

3. The method of claim 1 , wherein adaptively disabling the one or more filters comprises determining a value of a disabling flag that indicates whether the one or more filters are disabled.

4. The method of claim 3 , wherein:
the disabling flag is turned on if the syntax element is turned on, and
wherein if the disabling flag is on, the one or more filters are disabled.

5. The method of claim 3 , wherein:
the two-tap FIR filter is a first FIR filter, and
if the disabling flag is equal to 1 and a result of a second Finite Impulse Response (FIR) filter is determined to be above a threshold, the one or more filters are disabled.

6. The method of claim 1 , further comprising: summing residual blocks with corresponding predictive blocks to form a decoded video block, the corresponding predictive blocks including the predictive block.

7. The method of claim 1 , wherein interpolating the respective predicted sample by applying the bilinear interpolation filter comprises interpolating the respective predicted sample as follows:
predSamples[ x][y ]=((32 −i Fact)*ref[ x +iIdx +1 ]+i Fact*ref[ x+iIdx +2]+16)>>5

wherein ref[x+Idx+1] and ref[x+iIdx+2] are the references samples nearest to the sub-pixel location and iFact is a weighting between the reference samples nearest to the sub-pixel location.

8. A method of encoding video data, the method comprising:
adaptively disabling one or more filters used for intra prediction;
signaling a syntax element that controls the disabling of the one or more filters, the syntax element specifying whether residual signals are quantized without transformation; and
performing the intra prediction to generate a predictive block for a current video block of a current video slice of a current picture of the video data, wherein:
the one or more filters include at least one of a mode dependent intra smoothing (MDIS) filter, a bilinear interpolation filter, an edge filter, or a DC intra-prediction mode filter,
the MDIS filter is applied to reference samples differently for different block sizes and intra prediction modes,
when angular prediction is used to generate the predictive block, each respective predicted sample of the predictive block is obtained by projecting a location of the respective sample to a sub-pixel location between two of the reference samples nearest to the sub-pixel location and interpolating the respective predicted sample by applying the bilinear interpolation filter to interpolate at 1/32pixel accuracy using the two reference samples nearest to the sub-pixel location,
the edge filter filters a first prediction column of the predictive block and a first prediction row of the predictive block using gradient based smoothing, and
the DC intra-prediction mode filter filters the first prediction row of the predictive block and the first prediction column of the predictive block using a DC value with a two-tap Finite Impulse Response (FIR) filter and filters a corner sample of the predictive block using a 3-tap filter.

9. The method of claim 8 , wherein the syntax element is signaled in a prediction unit (PU).

10. The method of claim 8 , wherein adaptively disabling the one or more filters comprises determining a value of a disabling flag that indicates whether the one or more filters are disabled.

11. The method of claim 10 , wherein:
the disabling flag is turned on if the syntax element is turned on, and
if the disabling flag is on, the one or more filters are disabled.

12. The method of claim 10 , wherein:
the two-tap FIR filter is a first FIR filter, and
if the disabling flag is equal to 1 and a result of the first or a second Finite Impulse Response (FIR) filter is determined to be above a threshold, the one or more filters are disabled.

13. The method of claim 8 , wherein interpolating the respective predicted sample by applying the bilinear interpolation filter comprises interpolating the respective predicted sample as follows:
predSamples[ x][y ]=((32 −i Fact)*ref[ x +iIdx +1 ]+i Fact*ref[ x+iIdx +2]+16)>>5

wherein ref[x+Idx+1] and ref[x+iIdx+2] are the references samples nearest to the sub-pixel location and iFact is a weighting between the reference samples nearest to the sub-pixel location.

14. A video decoding device comprising:
a computer-readable medium configured to store video data, and
one or more processors configured to:
adaptively disable, based on a syntax element specifying whether residual signals are quantized without transformation, one or more filters used for intra-prediction; and

perform the intra prediction to generate a predictive block for a current block of a current video slice of a current picture of the video data based on a signaled intra prediction mode and data from previously decoded blocks of the current picture, wherein:
the one or more filters include at least one of a mode dependent intra smoothing (MDIS) filter, a bilinear interpolation filter, an edge filter, or a DC intra-prediction mode filter,
the MDIS filter is applied to reference samples differently for different block sizes and intra prediction modes,
when angular prediction is used to generate the predictive block, each respective predicted sample of the predictive block is obtained by projecting a location of the respective sample to a sub-pixel location between two of the reference samples nearest to the sub-pixel location and interpolating the respective predicted sample by applying the bilinear interpolation filter to interpolate at 1/32pixel accuracy using the two reference samples nearest to the sub-pixel location,
the edge filter filters a first prediction column of the predictive block and a first prediction row of the predictive block using gradient based smoothing, and
the DC intra-prediction mode filter filters the first pre

Assignees

• Qualcomm Inc

Inventors

• Joshi Rajan Laxman
• Sole Rojals Joel
• Karczewicz Marta
• Kang Jewon
• Kim Woo-Shik

Classifications

• H04N19/593 Primary

  involving spatial prediction techniques · CPC title

• H04N19/117

  Filters, e.g. for pre-processing or post-processing (sub-band filter banks H04N19/635) · CPC title

• H04N19/70

  characterised by syntax aspects related to video coding, e.g. related to compression standards · CPC title

• H04N19/176

  the region being a block, e.g. a macroblock · CPC title

• H04N19/82

  involving filtering within a prediction loop · CPC title