WO2013104198A1

WO2013104198A1 - Method and apparatus for unification of coefficient scan of 8x8 transform units in hevc

Info

Publication number: WO2013104198A1
Application number: PCT/CN2012/081628
Authority: WO
Inventors: Chih-Wei Hsu; Tzu-Der Chuang; Yu-Wen Huang
Original assignee: Mediatek Inc.
Priority date: 2012-01-13
Filing date: 2012-09-20
Publication date: 2013-07-18
Also published as: US10104399B2; EP2737707A1; US20150010072A1; AU2012365727B2; ZA201402544B; CN104041049A; CN104041049B; AU2012365727A1; EP2737707A4

Abstract

A method and apparatus for processing 2Nx2N transform units (TUs) are disclosed. In one embodiment, the method comprises determining a first-layer scanning order among four NxN sub-blocks of the 2Nx2N TU; determining a second-layer scanning pattern for said four NxN sub-blocks; and providing scanned 2Nx2N transform coefficients of the intra-coded or the inter- coded 2Nx2N TU using double scanning based on the first-layer scanning order and the second- layer scanning pattern. In another embodiment, said determining the first-layer scanning order is dependent on the second-layer scanning pattern. The second-layer scanning pattern can be diagonal, horizontal or vertical. In an embodiment, the first-layer scanning order can be from an upper-left sub-block, to an upper-right sub-block, to a lower-left sub-block and to a lower-right sub-block for the second-layer horizontal scanning pattern and from an upper-left sub-block, to a lower-left sub-block, to an upper-right sub-block and to a lower-right sub-block for other second-layer scanning patterns.

Description

METHOD AND APPARATUS FOR UNIFICATION OF COEFFICIENT SCAN OF 8X8 TRANSFORM. UNITS I.N HE VC

BACKGROUND OF THE INVENTION

Cross Reference To Related Applications

[0001] The present invention claims priority to U.S. Provisional Patent Application, Serial No. 61/586,248, filed on January 13, 2012, entitled "Unification of coefficient scan for 8x8 ΤΙΓ. The U.S. Provisional Patent Application is hereby incorporated by reference in its entirety.

Field of the Invention

[0002] The present invention relates to video coding, in particular, the present invention relates to coding techniques for coefficient scan of 8x8 transform units in High Efficiency Video Coding (HEVC).

Description of the Related Art

[0003] HEVC (High Efficiency Video Coding) is an advanced video coding system being developed under the Joint Collaborative Team on Video Coding (JCT-VC) group of video coding experts from ITU-T Study Group. In HEVC, a 2Nx2N coding unit (CU) can be hierarchically partitioned into a partition size selected from 2Nx2N, 2NxN, Nx2N and NxN. The current HEVC also defines the prediction unit (PU), where a CU may consist of one or multiple PUs. After prediction process is performed on each CU, the residues of each CU are further processed by transformation, in HEVC Test Model Version 5.0 (HM-5.0), transform coefficients of each transform unit (TU) are quantized and then scanned according to a certain scan pattern before Con lext-adaptive binary arithmetic coding (CABAC). For intra coded transform coefficients, the mode-dependent coefficient scan (MDCS) method is used. Three scan patterns, including diagonal J 10, horizontal 120, and vertical scanning 130, as illustrated in. Fig. 1, are used to scan the coefficients, A transform unit usually contains some zero- valued quantized transform coefficients in the high frequency region for typical image contents. The scanned coefficients may contain a string of zero- values data at the end. One effective way to code the scanned transform coefficients is to identify and code the string of zeros at the end of scanned transform coefficients. Therefore, the scanned coefficients are oflen examined from, the end toward the beginning to identify number of consecutive zero-valued data at the end of the scanned coefficients. Based on the number of consecutive zero-valued data at the end of the scanned transform coefficients, the consecutive zero-valued data can be coded very efficiently. The .number of consecutive zero- valued data at the end of the scanned transform coefficients can be identified by the last non-zero coefficient. In HEVC Test Model Version 5.0 ti-IM-5.0), the selection of the scan patterns for each intra-coded TU can be based on the intra prediction modes, as shown in Table 1 , where 0, 1, and 2 correspond to for diagonal, horizontal, and vertical scan patterns respectively. In HM-5.0, M.DCS is only used for intra TUs with sizes of 4x4 and 8x8 as shown in Table 1, while for inter-coded coefficients, only diagonal scanning is used for inter 4x4 and 8x8 TUs.

Table I .

|^'0004] As shown in Table I . diagonal scan pattern is applied to intra-coded 1.6x16 and 32x32 TUs, while different scan patterns are selectively applied to intra-coded 4x4 and 8x8 TUs.

[0005] After transform coefficients are re-ordered according to a scan pattern, CABAC compression is applied to the scanned transform coefficients on a TU by TU basis. In each TU, a last sigoificant eoeff X and a last significant cx>eff y are transmitted .first to indicate the last non-zero coefficient position in the predetermined scan order. The TU is then divided into several subsets when the TU size is larger than 4x4, The subsets in the TU are retrieved for coding operation one by one. For one 8x8 TU, the 64 coefficients are divided into 4 subsets according to the diagonal scan through the entire 8x8 TU. Each subset contains 16 continuous coefficients. Fig. 2 shows the 4 subsets in an 8x8 TU as indicated by four different shaded patterns. For square TUs larger than 8x8 (e.g. 1.6x16 and 32x32) and non-square TUs (e.g. 16x4, 4x 16. 32x8, and 8x32), the TU is divided into 4x4 sub-blocks. Each sub-block corresponds to a coefficient subset. For each subset, a significance map (significant ...coeff ...flag) indicating whether each coefficient is zero or not is coded first. Next, coeff abs level greaterl . flag, coeff_abs__level_greater2_flag, coeff_abs_level_minus3, and coetYj%ii_flag are utilized to represent each non-zero coefficient. In HM-5.0, if the TU size equals to 16x16, 32x32, 16x4. 4x 16, 3.2x8, or 8x3.2, one significant j;oeffgroup_ flag is coded for each sub-block prior to the level coding of the sub-block, (e.g. the significant coeff ...flag, coeff abs level, greaterl Jflag, coeff_absJevel_greater2_flag, coeff_abs_level_ininiis3, and coefT_sign_flag). if significant eoeffgroup ...flag equals to 0, it indicates that the entire 4x4 sub-block is zero. The level coding of the sub-block with zero for the entire 4x4 sub-block can be skipped. The significant _coef¾group ....flag is inferred as 3 for the DC sub-block.

[0006] In one TU, transform coefficients are divided into multiple subsets and similar CABAC coding operations are applied to each subset. However, in HM-5.0, there are two kinds of subset partitioning using diagonal scanning. One is to partition the scanned transform coefficients of the 8x8 TU using diagonal scanning into multiple subsets, and the other is to partition the scanned transform coefficients of the 8x8 TU into multiple subsets using double diagonal, scanning. In the latter case, each subset corresponds to a 4x4 sub-block. From implementation point of view, it is preferable to unify the subset partition methods in order to reduce system complexity and cost

BRIEF SUMMARY OF THE INVENTION

[0007] A method and apparatus for processing 2Nx2N transform units are disclosed. In one embodiment according to the present invention, the method for processing 2Nx2N transform units comprises receiving a 2Nx2.N TU (transform unit), determimng a -first-layer scanning order among four NxN sub-blocks of the 2Nx2N TU ; determining a second-layer scanning pattern for said four NxN sub-blocks; and providing scanned 2Nx2N transform coefficients for the 2Nx2N TU corresponds to the intra-coded TU or the inter-coded TU using double scanning based on the first-layer scanning order and the second-layer scanning pattern. The four NxN sub-blocks of the 2Nx2N TU are scanned based on the first-layer scanning order and each of said four NxN sub-blocks is scanned according to the second-layer scanning pattern.

[0008] The first-layer scanning order scans an upper left NxN sub-block first and scans a lower right NxN sub-block fast. Furthermore, in some embodiments of the present invention, said determining the first-layer scanning order can be dependent on the second-layer scanning pattern. The second-layer scanning pattern is selected from a second group consisting of diagonal scanning, horizontal scanning and vertical scanning. The first-layer scanning order can be from an upper- left sub-block, to an upper-right sub-block, to a lower-left sub-block and to a lower-right sub-block if the second-layer scanning pa ttern is the horizontal scanning. The first- layer scanning order can be from an upper-left sub-block, to a lower-left sub-block, to an upper- right sub-block and to a lower-right sub-block if the second- layer scanning pattern is the diagonal scanning or the vertical scanning. In yet another embodiment of the present invention, said determining the second-layer scanning pattern is dependent on an intra prediction mode associated with the 2Nx2N TU if the 2Nx2N TU is the intra-coded TU.

BRIEF DESCRIPTION OF DRAWINGS

[0009] Fig. I illustrates mode-dependent coefficient scan patterns for an intra-coded 8x8 transform unit in HM-5.0, where diagonal, horizontal and vertical scanning patterns are shown.

[0010] Fig. 2 illustrates the four subsets in an inter-coded 8x8 transform unit in HM-5.0.

[00.1.1] Fig. 3 illustrates two examples of .first- layer scan orders, where the first-layer scan order is used as the processing order of the four 4x4 sub-blocks of an 8x8 transform unit.

[0012] Fig. 4 illustrates three examples of two-layer scanning for intra-coded or inter-coded 8x8 transform units according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

[0013] As mentioned previously, for an intra-coded 8x8 TU in HM-5,0, the coefficients scanning is applied to the entire 8x8 TU using one of three possible scanning patterns according to intra prediction modes, and for an inter-coded 8x8 TU, diagonal scanning is applied to the entire 8x8 TU. An embodiment according to the present invention unifies the coefficient scan for 8x8 TUs with larger TU size and non-square TUs. Therefore, instead of using diagonal, horizontal, or vertical scanning for the entire 8x8 TU, an 8x8 TU is also partitioned into four 4x4 sub-blocks. Double scanning is used for the intra-coded or inter-coded 8x8 TU, Accordingly, 4x4 sub-blocks are used as subsets for coefficient scanning of intra-coded 8x8 TU and inter-coded 8x8 TU. An example of double diagonal scanning for inter-coded and intra- coded 8x8 TU is shown as pattern 31.0 in Fig. 3. When MDCS is used, there are two additional scan patterns, including horizontal and vertical scanning, for the entire 4x4 and 8x8 TUs. In another embodiment of the present invention, the horizontal scan pattern and die vertical scan pattern are also used for the 4x4 sub-blocks. Accordingly, the 8x8 TU is divided into four 4x4 sub-blocks and double horizontal scanning 320 or double vertical scanning 330 can be used as shown in Fig, 3.

[0014] Each of the double scanning patterns shown in Fig. 3 for the 8x8 TU is formed by applying two-layer scanning using the same scanning pattern, where the first-layer scanning is associated with the scanning order of the four 4x4 sub-blocks di vided from the 8x8 TU and the second- layer scanning is associated with the scan pattern within each 4x4 sub-block. For example, pattern 310 of Fig. 3 illustrates an example of applying the diagonal scanning pattern across the four sub-blocks and within each of the 4x4 sub-block. As shown in pattern 31.0 of Fig. 3, each of the 4x4 sub-blocks uses the diagonal scan pattern. Furthermore, diagonal scanning is also used as processing order across the four 4x4 sub-blocks. Accordingly, the upper-left sub-block is scanned first, and followed by lower-left, upper-right and lower-right sub-blocks. In pattern 320 of Fig. 3, each of the 4x4 sub-blocks uses the horizontal scan pattern. Furthermore, horizontal scanning is also used as processing order across the four 4x4 sub-blocks. Accordingly, the upper-left sub-block is scanned first, and followed by upper-right, lower-left and lower-right sub-blocks. Similarly, pattern 330 of Fig. 3 illustrates an example of vertical scanning across the four 4x4 sub-blocks and within each of the four 4x4 sub-blocks.

[0015] Fig. 3 illustrates examples of two layer scanning patterns using the same scanning pattera for both layers. The first layer scanning pattern involves the processing order across the four 4x4 sub-blocks, i.e., four 4x4 TUs. It is noted that the four 4x4 sub-blocks in a 8x8 TU are arranged, in a 2x2 shape. In the examples of Fig, 3, the first 4x4 scanning starts from the upper- left 4x4 sub-block, and ends at the lower-right comer 4x4 sub-block. While the scan order (i.e., the first-layer scanning) is always the same as the scanning pattern (i.e., the second-layer scanning) of the underlying 4x4 sub-blocks, the scanning for the two layers may also be selected independently. In Fig. 4, each box of the 2x2 boxes 410 and 420 represents a 4x4 sub-block. A vertical scan order is shown for 2x2 boxes 410 and a horizontal scan order is shown for 2x2 boxes 420. it is noted that the diagonal scan for 2x2 boxes is the same as the vertical scan in this case. The diagonal scan order and vertical scan order for the first layer will be different for square TUs with size greater than 8x8 and for certain non-square TUs (e.g. 32x8 TU). For some other non-square TUs (e.g. 4x16 TU and 16x4 TU), diagonal, horizontal, and vertical scan order applied in the first layer are all the same, hi some embodiments of the present invention, the first-layer scanning is selected independently from the second- layer scanning. For example, the vertical scanning order 410 can be used while each underlying 4x4 sub-block is scanned using horizontal scanning. All examples mentioned above always use the same scanning pattern for the four 4x4 sub-blocks. However, the four 4x4 sub-blocks may use different scanning patterns,

[0016] In the above example, an 8x8 TU is used to illustrate two-layer coefficient scanning, where the 8x8 TU is divided into four 4x4 sub-blocks. However, the present invention is not limited to the 8x8 TU. A skilled person in the art may practice the present, invention for 16x16 or 32x32 TUs or non-square TUs.

[00.17] The above description is presented to enable a person of ordinary skill in. the art to practice the present invention as provided in the context of a particular application and its requirement. Various modifications to the described embodiments will be apparent to those with skill in the art, and the general principles defined herein may be applied to other embodiments. Therefore, the present invention is not intended to be limited to the particular embodiments shown and described, but is to be accorded the widest scope consistent with the principles and novel features herein disclosed. In the above detailed description, various specific details are illustrated in order to provide a thorough understanding of the present invention. Nevertheless, it will be understood by those skilled in the art that the present invention may be practiced.

1003.8] Embodiment oi the present invention as described above may be implemented in various hardware, software code, or a combination of both. For example, an embodiment of the present invention can be a circuit integrated into a video compression chip or program code integrated into video compression software to perform the processing described herein. An embodiment of the present invention may also be program code to be executed on a Digital Signal Processor (DSP) to perform the processing described herein. The invention may also involve a number of functions io be performed by a computer processor, a digital signal processor, a microprocessor, or field programmable gate array (FPGA). These processors can be configured to perform particular tasks according to the invention, by executing machine-readable software code or firmware code that defines the particular methods embodied by the invention. The software code or firmware code may be developed in different programming languages and different formats or styles. The software code may also be compiled for different target platforms. However, different code formats, styles and languages of software code and other means of configuring code to perform the tasks in accordance with the invention will not. depart from the spirit and scope of the invention.

[0019] The invention may be embodied in other specific forms without departing from its spirit or essential characteristics. The described examples are to be considered in all respects only as illustrative and not restrictive. The scope of the invention is therefore, indicated by the appended claims rather than by the foregoing description. Ail changes which come within the meaning and range of equivalency of the claims are to be embraced within their scope.

Claims

1 . A method for processing 2Nx2N transform units, the method comprising:

receiving a 2Nx2N TU (transform unit), where the 2Nx2N transform unit corresponds to an intra-coded TU or an inter-coded TU, and wherein N is an integer from a group consisting of 2,

4, 8, or 16;

determining a first-layer scanning order among four NxN sub-blocks of the 2Nx2N TU, wherein said four NxN sub-blocks of the 2Nx2N TU are scanned based on the first-layer scanning order;

determining a second-layer scanning pattern for said four NxN sub-blocks, wherein each of said four NxN sub-blocks is scanned according to the second-layer scanning pattern; and

providing scanned 2Nx2N transform coefficients for the 2Nx2N TU corresponds to the intra-coded TU or the inter-coded TU using double scanning based on the first-layer scanning order and the second-layer scanning pattern.

2. The method of Claim L wherein the first-layer scanning order scans an upper left NxN sub-block first and scans a lower right NxN sub-block last.

3. The method of Claim 1 , wherein said determining the first-layer scanning order is dependent on the second- layer scanning pattern.

4 The method of Claim .1 , wherein the second-layer scanning pattern is selected from a group consisting of diagonal scanning, horizontal scanning and vertical scanning.

5, The method of Claim 4, wherein the first-layer scanning order is from an upper-left sub- block, to an upper-right sub-block, to a lower-left sub-block and to a lower-tight sub-block if ihe second-layer scanning pattern is the horizontal scanning.

6, The method of Claim 4, wherein the first-layer scanning order is from an upper-left sub- block, to a lower-left sub-block, to an upper-right sub-block and to a lower-right sub-block if the second-layer scanning pattern is the diagonal scanning or the vertical scanning.

7, The method of Claim 1 , wherein said determining the second-layer scanning partem is dependent on an intra prediction mode associated with the 2Nx2N TU if the 2Nx2N TU is the intra-coded TU.

8. An apparatus for processing 2Nx2N transform units, the apparatus comprising:

means for receiving a 2Nx2N TU (transform unit), where the 2Nx2N transform unit corresponds to an mtra-coded TU or an inter-coded TU, and wherein N is an integer from a group consisting of 2, 4, 8, or 16;

means for determining a first-layer scanning order among four NxN sab-blocks of the 2Nx2N TU. wherein said four NxN sub-blocks of the 2Nx2N TU are scanned based on the first- layer scanning order;

means for determining a second-layer scanning pattern for said four NxN sub-blocks, wherein each of said four NxN sub-blocks is scanned according to the second-layer scanning pattern; and

means for providing scanned 2Nx2N transform coefficients for the 2Nx2N TU corresponds to the intra-coded TU or the inter-coded TU using double scanning based on the first-layer scanning order and the second-layer scanning pattern.

9. The apparatus of Claim 8, wherein the first-layer scanning order scans an upper left NxN sub-block first and scans a lower right NxN sub-block last.

10. The apparatus of Claim 8, wherein said means for said determining the first-layer scanning order is dependent on the second-layer scanning pattern.

1 i. The apparatus of Claim 8, wherein, the second-layer scanning pattern is selected from a second group consisting of diagonal scanning, horizontal scanning and vertical scanning.

12. The apparatus of Claim 1 1 , wherein the first-layer scanning order is from an upper-left sub-block, to an upper-right sub-block, to a lower-left sub-block and to a lower-right sub-block if the second-layer scanning pattern is the horizontal scantling.

13. The apparatus of Claim 1 1 , wherein the first-layer scanning order is from an upper-left sub-block, to a lower-left sub-block, to an upper-right sub-block and to a lower-right sub-block if the second-layer scanning pattern is the diagonal scanning or the vertical scanning.

14. The apparatus of Claim 8, wherein said means for said determining the second-layer scanning pattern is dependent on an intra prediction mode associated with the 2Nx2N TU if the 2Nx2N TU is the intra-coded TU.

15. A method for processing transform units, the method comprising:

receiving a TU (transform unit) comprising of a plurality of 4x4 sub-blocks;

determining a first-layer scanning order among the plurality of 4x4 sub-biocks of the TU, wherein the plurality of 4x4 sub-blocks of the TU are scanned based on the first-layer scanning order;

determining a second-layer scanning pattern for scanning transform coefficients in the plurality of 4x4 sub-blocks, wherein each 4x4 sub-block is scanned according to the second- layer scanning pattern, and the second-layer scanning pattern is horizontal scanning or vertical scanning; and

providing scanned transform coefficients for the TU using double scanning based on the first- layer scanning order and the second-layer scanning pattern.

16. The method of Claim 15, wherein the first-layer scanning order is selected from a group consisting of diagonal scanning, horizontal scanning and vertical scanning.

17. The method of Claim 15, wherein the TU is a non-square TU.