Low-complexity deblocking filter

• US 20040101059A1
• Filed: 12/05/2002
• Published: 05/27/2004
• Est. Priority Date: 11/21/2002
• Status: Active Grant

• Current Assignee: Cisco Systems Canada Co. (Cisco Systems, Inc.)
• Original Assignee: UB Video, Inc. (Cisco Systems, Inc.)
• Inventors: LIn, Yu-sheng Brandon, Joch, Anthony, Au, James
• Granted Patent: US 7,050,504 B2
• Time in Patent Office: Days
• US Class Current: 375/240.29
• CPC Class Codes: H04N 19/117 Filters, e.g. for pre-processing or post-processing sub-band filter banks H04N19/635
• CPC Class Codes: H04N 19/14 Coding unit complexity, e.g. amount of activity or edge presence estimation H04N19/146 takes precedence
• CPC Class Codes: H04N 19/159 Prediction type, e.g. intra-frame, inter-frame or bidirectional frame prediction
• CPC Class Codes: H04N 19/176 the region being a block, e.g. a macroblock
• CPC Class Codes: H04N 19/86 involving reduction of coding artifacts, e.g. of blockiness

Abstract

A method of filtering to remove coding artifacts introduced at block edges in a block-based video coder, the method having the steps of: checking the content activity on every line of samples belonging to a boundary to be filtered and where content activity is based on a set of adaptively selected thresholds determined using Variable-Shift Table Indexing (VSTI); determining whether the filtering process will modify the sample values on that particular line based on said content activity, and selecting a filtering mode between at least two filtering modes to apply on a block boundary basis, implying that there would be no switching between the two primary modes on a line by line basis along a given block boundary. The two filtering modes include a default mode based on a non-recursive filter, and a strong filtering mode which features two strong filtering sub-modes and a new selection criterion that is one-sided with respect to the block boundary to determine which of the two strong filtering sub-modes to use. The two strong filtering sub-modes include a new 3-tap filter sub-mode and a 5-tap filter sub-mode that permits a more efficient implementation of the filter.

Claims

1. A method of filtering samples to minimise coding artifacts introduced at a block boundary in a block-based video encoder, the method having the steps of:

(a) calculating a pair of indices used to access a table of a pair of corresponding activity threshold values, the indices calculated using an average quantization parameter and an offset parameter;

(b) determining the activity threshold values based on the pair of indicies;

(c) confirming whether the filtering process will modify the sample values on every line of samples for the block boundary by checking a content activity for the every line of samples for said block boundary, the content activity based on the determined activity threshold values; and

(d) filtering the confirmed samples when a block on either side of the block boundary was coded using inter prediction.

2. The method of claim 1, wherein a pair of the activity threshold values are ALPHA (α

) and BETA (β

), and the average quantization parameter, QP_av, was used in coding the two blocks that neighbour the block boundary.

3. The method of claim 2, wherein the values of the activity threshold values, ALPHA(α

) and BETA (β

) for each possible value of the indices Index_A and Index_B are

4. The method of claim 1, wherein the following equations are used to compute the filtered sample values at a second position from the block boundary on either side, denoted P₁ and Q₁, for a luminance component:

P₁=p₁+Clip(−

C0, C0, (p₂+(p₀+q₀)>

>

1−

(p₁<

<

1))>

>

1) Q₁=q₁+Clip(−

C0, C0, (q₂+(p₀+q₀)>

>

1−

(q₁<

<

1))>

>

1)

5. The method of claim 4, wherein the maximum change in the value of the samples that neighbour the block boundary, p₀ and q₀, due to the filtering process for chrominance components is computed as

C=C0+1.

6. A method of filtering according to claim 2, wherein a pair of the offset parameter is added to the average QP value, QP_av, in order to compute the indices that are used to address the tables of ALPHA (α

) and BETA (β

) activity threshold values, such that

Index_A=Clip(QP_min, QP_max, QP_av+Filter_Offset_A) Index_B=Clip(QP_min, QP_max, QP_av+Filter_Offset_B)

7. A method of filtering according to claim 4, wherein the value of the C0 parameter used for clipping the change in the sample values is computed using a value of Index_A of the indicies to address the table of C0 values.

8. A method of controlling filter properties to adjust the properties of said filter at a block boundary, said method having the steps of:

(a) computing an average quantization parameter value (QP_av) at said block boundary,

(b) adding offset values Filter_Offset_A and Filter_Offset_B to the average quantization parameter value QP_av and clipping these values within a given range to determine table indices Index_A and Index_B; and

(c) accessing an ALPHA (α

) table, a BETA (β

) table, and a Clipping (C0) table using the indices computed based on the filter offsets and the average quantization parameter value such that;

ALPHA=ALPHA_TABLE[Index_A]

BETA=BETA_TABLE [Index_B]

C0=CLIP_TABLE[BS][Index_A]

9. A method of filtering samples to minimise coding artifacts introduced at a block boundary in a block-based video encoder, the method having the steps of checking content activity on every line of samples belonging to the boundary to be filtered and determining whether the filtering process will modify the sample values on said line of samples based on content activity thresholds that are dependent on a quantization parameter and determined using a filter offset parameter.