CN108449602A - Method and apparatus for encoding coded block flag and decoding video bitstream - Google Patents

Abstract

The invention discloses a method and a device for coding a coding block flag and decoding a video bit stream. The method for decoding the video bitstream includes: receiving the video bitstream from a media or a processor; according to a first context formation, a decoding process based on context-adaptive binary arithmetic coding restores a first coding block flag from the video bitstream; and restoring a second coded block flag from the video bitstream based on the decoding process of context adaptive binary arithmetic coding according to a second context formation, wherein the first context formation and the second context formation both depend on a depth of a residual quadtree. The method and the device can simplify the signaling method of the flag of the coding block and improve the efficiency.

Description

Method and device for encoding coded block flag and decoding video bitstream

technical field

The present invention relates to video coding, and in particular, the present invention relates to coding block flags associated with coding units (CU) and transform units (TU) in High-Efficiency Video Coding (HEVC) A coding method and device for a coded block flag (cbf) syntax (syntax) and a decoding method and device for a video bit stream.

Background technique

HEVC is an advanced video coding system developed by the Joint Collaborative Team on Video Coding (JCT-VC for short) composed of video coding experts from the ITU-T research group. In HEVC test version 5.0 (HM-5.0), inter-coded (inter-coded) and intra-coded (intra-coded) residuals are encoded using block-based transform coding. The above blocks (referred to as TUs) are partitioned from the root block (root TU) using a quadtree structure. The quadtree partitioning described above may be applied recursively until a leaf block or minimum block is produced. Afterwards, a 2D transformation is applied to each transformation unit. Each transform unit (TU) may be divided into four sub-TUs, ie, leaf-TUs. For each TU, a syntax element called cbf (coded block flag) needs to be transmitted to indicate whether the TU has non-zero transform coefficients, where "1" means that there is at least one non-zero transform coefficient, and "0" means There are no non-zero transform coefficients.

In HM-5.0, for the luminance component, cbf will only signal the leaf TU of the residual quadtree. For the chroma component, cbf sends a signal to both the root TU and the leaf TU. However, the cbf only sends a signal in a TU whose size is smaller than or equal to the maximum chroma TU size. Figures 1 to 3 show examples of cbf signaling. In FIG. 1 , block 110 shows the residual quadtree partitioning of TUs, where the root TU is partitioned into sub-TUs (TU0-TU 6) using quadtree partitioning. Block 120 shows the corresponding cbf bits where TU 1 , TU 3 , TU 5 and TU 6 have non-zero transform coefficients, while TU0 , TU 2 and TU 4 do not have non-zero transform coefficients. If the TU is a luma TU, the cbf bit is only sent for the leaf TU. 2A is an example of cbf signaling (cbf encoding) for luma TUs, where four groups of bits "0", "1", "0101" and "1" correspond to the four leaf-TUs of the root TU 210, respectively. cbf bits. The above cbf bits are transmitted in the order of raster-scan, that is, in the order of upper left TU, upper right TU, lower left TU, and lower right TU. For the lower left leaf TU, it is further divided into 4 leaf TUs. In sequential scan order, the cbf bits of this leaf-TU are "0101". Accordingly, FIG. 2A shows four groupings of cbf bits 220 . Figure 2B shows an example of cbf signaling for chroma TUs, where cbf bits are transmitted for root TU and leaf TU. The root TU 230 is partitioned into four leaf-TUs, and the lower left leaf-TU is further partitioned into four leaf-TUs. Therefore, there are three levels of cbf bits corresponding to the three levels of the TU. For the root TU (ie, depth=0), the cbf bit "1" (represented by number 240) is signaled. For the four leaf-TUs of the root TU, the cbf bits are "0", "1", "1" and "1" (represented by number 250), respectively, in sequential scan order. For the lower left leaf TU, it is further divided into four leaf TUs. According to the sequential scanning order, the above four leaf TUs correspond to cbf bits "0", "1", "0" and "1" respectively (by number 260). As shown in FIG. 2A and FIG. 2B , although the luma TU and the chroma TU use the same RQT (residual quadtree), the signaling of cbf is different. The example shown in FIG. 2B is applicable to the case where the root block size is smaller than or equal to the maximum chroma TU size. For example, assuming the maximum chroma TU size is 16x16 and the minimum chroma TU size is 4x4, the root TU 230 is 16x16 in size, and each leaf TU in each lower left leaf TU is of size 4 ×4. When the size of the chroma leaf TU is larger than the maximum chroma TU size, for example, 32×32, cbf signaling will not be performed at the 32×32 level.

In order to reduce the number of cbf bits, an inference method can be applied to luma and chroma TUs, in which the cbf of the fourth leaf TU of the root TU can be inferred using the cbf of other TUs. Therefore, there is no need to send a signal to the cbf of the fourth leaf TU.

For a luma TU, the cbf of the fourth leaf TU can be inferred from the cbf of the previous three leaf TUs and the cbf of the associated root TU. Block 310 in FIG. 3 shows an example of how the cbf of the fourth leaf-TU can be inferred. The lower left TU indicated by the thick line box 312 is divided into four leaf-TUs, where the cbf of the fourth leaf-TU is 1. Since the TU 312 is partitioned into four leaf-TUs, there is at least one non-zero transform coefficient in the four leaf-TUs. When the cbfs of the three previous leaf-TUs are all 0 (in sequential scan order), the cbf of the last leaf-TU (ie, the fourth leaf-TU) must be 1. Therefore, in this case, the cbf of the fourth leaf-TU can be inferred. For brevity, the cbf of a leaf-TU is also referred to as the leaf-cbf.

For chroma TUs, it is different because cbf transmission needs to be done for all residual quadtree levels. For the four leaf-TUs associated with each root TU, the cbf of the root TU is transmitted. If the cbf of a TU (block 312 in FIG. 3 ) is 1, there must be at least one non-zero leaf-TU among the four leaf-TUs. Therefore, if the cbfs of the first three leaf-TUs are all 0, then the cbf of the last leaf-TU (indicated by the circle) must be 1. In this case, the last cbf can be deduced without being signaled. Furthermore, both the intra- and inter-coded TUs of the chroma components can apply the above inference mechanism.

In HEVC, there is a root residual flag for inter-coded CUs. When the residual flag is false, it is not necessary to send all the cbf of the Y, U, V components. When the residual flag is true and the TU depth of the current CU is 0, it can be inferred that the luma cbf is 1 if both the chroma cbf are 0. Therefore, if the cbf of U (block 320 ) and V (block 330 ) are both 0, the cbf of a luma TU at depth 0 can be inferred to be 1, as shown in FIG. 3 .

In HM-5.0, the maximum TU sizes for chroma and luma components are 16×16 and 32×32, respectively. However, the maximum CU size for chroma components is 32×32. Therefore, its maximum CU size and maximum TU size are not the same. Furthermore, in HM-5.0, the chroma cbf only sends signals for TUs whose size is smaller than or equal to the maximum TU size. For example, when the CU size is 64×64, ie, the chroma CU size is 32×32, the maximum TU size corresponds to 16×16. Therefore, four root cbfs will be transmitted for four 16x16 chroma TUs in a 32x32 CU. In this case, even if all four cbfs are 0, the cbfs will be transmitted, as shown in FIG. 4 , where the size of the chroma CU 410 is 32×32.

As mentioned above, the cbf signaling method for luma TU and chroma TU is different. Therefore, a unified cbf signaling method is needed to simplify the above process. In addition, there are some redundancies in the existing cbf signaling method, so it is also necessary to further improve the efficiency of the existing cbf signaling method.

Contents of the invention

In view of this, the following technical solutions are provided:

Embodiments of the present invention provide a decoding method and device for a video bitstream, wherein the method includes: receiving the video bitstream from a media or a processor; forming a decoding process based on context-adaptive binary arithmetic coding from the first context Restoring a first coded block flag in the video bitstream, wherein the first coded block flag is associated with a first transform unit of a first color component, the first coded block flag indicates the first transform of the first color component whether the unit has at least one non-zero transform coefficient; and according to a second context formation, a decoding process based on the context-adaptive binary arithmetic coding restores a second coded block flag from the video bitstream, wherein the second coded block flag Associated with a second TU of a second color component, the second coding block flag indicates whether the second TU of the second color component has at least one non-zero transform coefficient, wherein the first color component and the second The color components are different, and both the first context formation and the second context formation depend on the depth of the residual quadtree.

Embodiments of the present invention provide an encoding method and device for encoding a block flag, wherein the method includes: receiving a first transformation unit of a first color component and a second transformation unit of a second color component from a medium or a processor; A first residual quadtree related to the first TU and a second residual quadtree related to the second TU; determine a first coding block flag corresponding to the first TU and a flag corresponding to the first TU The second coding block flag of two transformation units, wherein the first coding block flag indicates whether the first transformation unit has at least one non-zero transform coefficient, and the second coding block flag indicates whether the second transformation unit has at least one a non-zero transform coefficient; and a coding process based on context-adaptive binary arithmetic coding encoding the first coded block flag formed according to a first context and formed according to a second context based on the context-adaptive binary arithmetic coding An encoding process encodes the second coded block flag to generate a video bitstream in which the first color component is different from the second color component and both the first context formation and the second context formation depend on residual The depth of the quadtree.

The method and device described above can simplify the signaling method of the coding block flag and improve the efficiency.

Description of drawings

FIG. 1 is a diagram illustrating an example of a residual quadtree structure and coded block flags of a leaf TU.

FIG. 2A is a diagram illustrating an example of a coding block flag signaling method for a luma TU according to HM-5.0.

FIG. 2B is a diagram illustrating an example of a coding block flag signaling method for a chroma TU according to HM-5.0.

FIG. 3 is a diagram illustrating an example of coding block flag signaling based on inferred luma TUs and chroma TUs.

FIG. 4 is a diagram illustrating an example of CBF signaling for CBFs of four 16×16 chroma root transform units.

FIG. 5 is a diagram illustrating an example of a coded block flag inference mechanism for an Inter CU according to an embodiment of the invention.

6A and 6B are diagrams illustrating an example of signaling of coding block flags of chroma components at CU level according to an embodiment of the present invention.

FIG. 7 is a schematic diagram of an exemplary flow of an encoder integrating an embodiment of the present invention.

FIG. 8 is a schematic diagram of an exemplary process of integrating a decoder according to an embodiment of the present invention.

FIG. 9 is a schematic diagram of an exemplary process of integrating an encoder according to another embodiment of the present invention.

FIG. 10 is a schematic diagram of an exemplary process of integrating a decoder according to another embodiment of the present invention.

Detailed ways

In one embodiment of the present invention, by extending the chroma cbf encoding method to luma cbf, the signaling methods of luma and chroma cbf are unified. Therefore, both luma and chrominance cbf are sent for each level of the residual quadtree. In other words, cbf sending will be carried out for root TU and leaf TU. In this case, the inference methods for luma and chroma components are also unified. Therefore, luma TU and chroma TU use the same inference method. In other words, if the cbf of the first three leaf-TUs are all 0, then the cbf of the last leaf-TU must be 1.

In another embodiment, the residual flag derivation method for Inter CUs is applied to the unified signaling method. Therefore, when the residual flag is true and the cbf of the chroma TU is all 0, the cbf of the top-level root luma TU is inferred to be 1 regardless of whether the top-level root TU can be further partitioned. Furthermore, the above-mentioned residual flag inference method for inter CUs can also be applied to other TU depths except depth 0. In other words, when the TU is further partitioned and the chroma cbfs are all 0, the cbf of the luma TU can be inferred to be 1. As shown in FIG. 5 , when the residual flag is 1 and the cbfs for the chroma root TUs (U 520 and V 530 ) are both 0, the cbf of the luma root TU 510 may be inferred to be 1.

Furthermore, the context formation of chroma cbf can also be unified with chroma cbf, so that the context formation of cbf coding based on context-based adaptive binary arithmetic coding (CABAC) is more effective for luma Both depend on TU depth for chroma components. To reduce the complexity of entropy coding of cbf flags, the number of contexts can be reduced. In addition, a bypass coding mode is also available for CABAC-based cbf coding.

In another embodiment, the root cbf is always signaled at the CU level regardless of the maximum TU size. Therefore, there is always a root cbf in each CU. 6A and 6B are examples of the cbf encoding process when the chroma CU size is 32x32 and the maximum TU size is 16x16. In Figure 6A, a chroma CU corresponds to a 32x32 block, which is larger than the maximum chroma TU size (ie, 16x16). Since all chroma TUs associated with a CU have no non-zero transform coefficients (represented by 0), the root cbf for a chroma CU is 0. According to an embodiment of the present invention, the root cbf in each CU is always signaled, so 0 will be signaled for the CU without other cbf signaling. Another example is shown in Fig. 6B, where the bottom left TU contains at least one non-zero transform coefficient. In this case, 1 is sent for the root chroma CU, and additional cbf bits "0 0 1 0" are also signaled to indicate which TU contains non-zero transform coefficients. In coding systems based on HM-5.0, the maximum TU sizes for luma and chrominance components are known. Information about the maximum TU size can also be signaled in the bitstream, for example at the sequence level of the bitstream (eg, SPS).

In yet another embodiment, by extending the encoding method of luma cbf to chroma cbf, the signaling methods of luma and chroma cbf are unified. Thus, both luma and chroma cbfs are signaled only for leaf-TUs.

The cbf signaling method described above can be applied to a video encoder, and can also be applied to a video decoder. FIG. 7 is a schematic diagram of an exemplary flow of an encoder integrating an embodiment of the present invention. In step 710, the residual of the current CU is determined, wherein the size of the current CU is larger than the maximum TU size. In step 720, a first cbf of the color component is determined, the first cbf indicates whether the current CU (depth=0) has at least one non-zero transform coefficient. As shown in step 730, different processing paths may be selected according to the result of the first cbf. If the current CU of the color component has at least one non-zero transform coefficient, then in step 740, four second cbfs of the color component are determined, wherein each second cbf indicates four sub-blocks (depth=1) of the color component in the current CU ) has at least one non-zero transform coefficient. In this case, as shown in step 750, the first cbf and the four second cbfs will be integrated into the video bitstream. If the current CU does not have non-zero transform coefficients, as shown in step 760, only the first cbf is integrated into the video bitstream. The cbf signaling by integrating the cbf into the video bitstream will allow the decoder to restore the residual quadtree structure and execute the corresponding decoding procedure. In some embodiments, if at least one sub-block of a color component has at least one non-zero transform coefficient, and the sub-block does not reach the minimum TU size of the color component, the sub-block with non-zero transform coefficient is further divided into 4 leaf blocks (depth=2). For each sub-block with non-zero transform coefficients, four third cbfs are determined, where each third cbf indicates whether one of the four leaf blocks of the color component has at least one non-zero transform coefficient. The four third cbfs of the above color components are also integrated into the video bitstream. The aforementioned sub-blocks and leaf blocks may be root TUs and leaf-TUs in the current CU. The aforementioned color components may be luma or chrominance components.

FIG. 8 is a schematic diagram of an exemplary process of integrating a decoder according to an embodiment of the present invention. In step 810, a video bitstream is received from a media or processor. The above-mentioned video bitstream may be stored in a medium, such as a storage medium (hard disk, optical disk, flash memory card), or computer memory (RAM, PROM, DRAM or flash memory). A video bitstream may also be received and/or processed by the processor. For example, in a broadcast environment, a channel receiver may receive the modulated signal, demodulate and demultiplex to recover the desired bit stream. In this case, the video bitstream is received from the processor (ie, channel receiver). In step 830, the first cbf of the color component is decoded, the first cbf indicates whether the current CU (depth=0) of this color component has at least one non-zero transform coefficient. As shown in step 840, different decoding paths may be selected according to the decoding result. If the first cbf of the color component is not 0, four second cbfs of the color component are decoded in step 850, wherein each second cbf indicates one of the four sub-blocks (depth=1) of the color component in the current CU has at least one non-zero transform coefficient. In step 860, the residual quadtree structure of the current CU of the color component is determined based on the first cbf and the four second cbfs. If the four second cbfs of the color component are all 0, then as shown in step 870, the residual quadtree structure of the current CU of the color component is determined based only on the first cbf. In some embodiments, if the color component has at least one non-zero transform coefficient in a sub-block of depth=1, and the size of this sub-block is larger than the minimum TU size of the color component, four third cbfs of the color component are decoded. Each third cbf of the color component indicates whether one of the four leaf blocks of the color component has at least one non-zero transform coefficient. At this time, the residual quadtree structure of the current CU of the color component is determined based on the above-mentioned first cbf, second cbf and third cbf. The aforementioned sub-blocks and leaf blocks may be root TUs and leaf-TUs in the current CU. The aforementioned color components may be luma or chrominance components.

FIG. 9 is a schematic diagram of an exemplary process of integrating an encoder according to another embodiment of the present invention. In step 910, a TU is received from a medium or a processor. Then in step 920 the RQT (residual quadtree) associated with this TU is determined. In step 930, one or more cbfs corresponding to the RQT of the TU are determined, wherein the signaling of the cbfs for luma and chrominance components is the same.

FIG. 10 is a schematic diagram of an exemplary process of integrating a decoder according to another embodiment of the present invention. In step 1010, a video bitstream is received from a media or processor. In step 1020, the cbf associated with the TU is determined, wherein the cbf is recovered from the video bitstream. In step 1030, the residual quadtree related to the TU is determined based on the above cbf, wherein the signaling of the cbf for the luma component and the chrominance component is the same.

The above-mentioned process is only for describing an example of cbf signaling for the encoder and decoder integrating the embodiment of the present invention. Without departing from the spirit of the present invention, those skilled in the art can modify each step, reorder the above steps, split one step, or integrate multiple steps to realize the present invention.

The above description is to enable those skilled in the art to implement the present invention with specific implementations and requirements in the context. However, those skilled in the art can make various variations and modifications, and the basic spirit of the present invention can also be applied to other embodiments. Therefore, the present invention is not limited to the specific embodiments described above, but should be defined in the broadest scope consistent with the spirit and new features of the present invention. In the above detailed description, various specific details are set forth to facilitate a comprehensive understanding of the present invention, however, those skilled in the art should understand that the present invention may be implemented without limiting some or all of these specific details.

The above-mentioned embodiments according to the present invention can be implemented by different hardware, software codes, or a combination of both. For example, according to an embodiment of the present invention, it may be a circuit integrated into a video compression chip for implementing the method, or a program code integrated into video compression software. According to another embodiment of the present invention, it may also be a program code executed on a digital signal processor (Digital Signal Processor, DSP) to implement the method. The present invention may also include a series of functions performed by a computer processor, DSP, microprocessor, or field programmable gate array (Field Programmable Gate Array, FPGA). In accordance with the present invention, these processors can be configured to perform specific tasks by executing machine-readable software code or firmware code that defines specific methods of embodiments of the present invention. The software code or firmware code may be developed in different programming languages and in different formats/styles. The software code can also conform to different target platforms. However, software codes with different code formats, styles and languages, and codes formed in other ways to perform the tasks corresponding to the present invention should be included within the scope of the present invention.

The present invention can also be implemented in other specific forms without departing from the spirit and essential features of the present invention. The above-mentioned embodiments are only for illustrating the present invention, and are not intended to limit the present invention. The scope of the present invention shall be determined by the appended claims, and all equivalent changes and modifications made according to the claims of the present invention shall fall within the scope of the present invention.

Claims (14)

1. a kind of coding/decoding method of video bit stream, including：

The video bit stream is received from media or processor；

It is formed according to the first context, based on the decoding process of context adaptive binary arithmetic coding from the video bit stream The first encoding block flag of middle reduction, wherein the first encoding block flag is related to the first converter unit of the first color component, should First encoding block flag indicates whether first converter unit of first color component has an at least non-zero transform coefficient；With And

It is formed according to the second context, based on the decoding process of the context adaptive binary arithmetic coding from the video bits The second encoding block flag is restored in stream, wherein the second encoding block flag is related to the second converter unit of the second color component, The second encoding block flag indicates whether second converter unit of second color component has an at least non-zero transform coefficient,

Wherein first color component is different from second color component, and first context is formed and the second context shape At the depth for each depending on residual error quaternary tree.

2. the coding/decoding method of video bit stream as described in claim 1, which is characterized in that first color component is brightness point Amount, second color component are chromatic component.

3. the coding/decoding method of video bit stream as described in claim 1, which is characterized in that first context formation depends on With the depth of the relevant residual error quaternary tree of first converter unit.

4. the coding/decoding method of video bit stream as described in claim 1, which is characterized in that second context formation depends on With the depth of the relevant residual error quaternary tree of second converter unit.

5. the coding/decoding method of video bit stream as described in claim 1, which is characterized in that the first encoding block flag and this One in two encoding block flags transmits at root converter unit and leaf transformation unit, the first encoding block flag and second volume Another in code block flag is transmitted at leaf transformation unit and is not transmitted at root converter unit.

6. the coding/decoding method of video bit stream as described in claim 1, which is characterized in that no matter the block size of coding unit is No to be more than maximum converter unit size, at least one of the first encoding block flag and the second encoding block flag are in the coding The root rank of unit transmits.

7. a kind of coding method of encoding block flag, including：

The second transformation that the first converter unit and the second color component of the first color component are received from media or processor is single Member；

Determine with the relevant first residual error quaternary tree of first converter unit and with relevant second residual error of second converter unit Quaternary tree；

Determine the first encoding block flag corresponding to first converter unit and the second volume corresponding to second converter unit Code block flag, wherein the first encoding block flag indicate whether first converter unit has an at least non-zero transform coefficient, should Second encoding block flag indicates whether second converter unit has an at least non-zero transform coefficient；And

It is formed according to the first context, based on the coding pass of context adaptive binary arithmetic coding to first encoding block Flag is encoded, and is formed according to the second context, the coding pass based on the context adaptive binary arithmetic coding The second encoding block flag is encoded, to generate video bit stream,

Wherein first color component is different from second color component, and first context is formed and the second context shape At the depth for each depending on residual error quaternary tree.

8. the coding method of encoding block flag as claimed in claim 7, which is characterized in that first color component is brightness point Amount, second color component are chromatic component.

9. the coding method of encoding block flag as claimed in claim 7, which is characterized in that first context formation depends on With the depth of the relevant residual error quaternary tree of first converter unit.

10. the coding method of encoding block flag as claimed in claim 7, which is characterized in that second context formation is depended on In the depth with the relevant residual error quaternary tree of second converter unit.

11. the coding method of encoding block flag as claimed in claim 7, which is characterized in that the first encoding block flag and should One in second encoding block flag transmits at root converter unit and leaf transformation unit, the first encoding block flag and this second Another in encoding block flag is transmitted at leaf transformation unit and is not transmitted at root converter unit.

12. the coding method of encoding block flag as claimed in claim 7, which is characterized in that no matter the block size of coding unit Whether maximum converter unit size is more than, and at least one of the first encoding block flag and the second encoding block flag are in the volume The root rank of code unit transmits.

13. a kind of decoding apparatus of video bit stream, including：

Processor, in program stored in executing memory, perform claim requires the video bits described in any one of 1-6 The step of coding/decoding method of stream.

14. a kind of code device of encoding block flag, including：

Processor, in program stored in executing memory, perform claim requires the encoding block described in any one of 7-12 The step of coding method of flag.