CN112887734B - Method and device for transforming in video image encoding and decoding - Google Patents

Abstract

The present disclosure relates to a method for transformation in video coding and decoding, the method comprising: obtaining a residual of a current coding block according to a prediction value of a current coding block of an image and an original value of the image; dividing the current coding block into n sub-transformation blocks according to mode information for dividing the current coding block into sub-transformation blocks, and performing transformations corresponding to the sub-transformation blocks on the generated residuals of different sub-transformation blocks, wherein n is an integer greater than or equal to 3; quantizing the transformation result to obtain coefficients of the sub-transformation blocks, and performing coefficient coding on the coefficients of the sub-transformation blocks; and writing the coefficient coding result and the mode information for dividing the current coding block into sub-transformation blocks into a bitstream.

Description

Method and device for transforming in video image encoding and decoding

Technical Field

The present invention relates to the technical field of video coding/decoding, and more specifically, to a method and device for conversion in coding/decoding. The present invention relates to the technical field of video coding/decoding, and more specifically, to a method and device for conversion in coding/decoding.

Background Art

In natural videos, there is a high degree of similarity between images. Therefore, in order to remove information redundancy between images during video compression, inter-frame prediction is often performed. The so-called inter-frame prediction is to use the already encoded image to predict the current image to be encoded, and then pass the prediction error to the decoder. Compared with directly transmitting the content of the current image, the amount of information in the prediction error is much smaller, thus achieving the purpose of compression. In fact, when performing inter-frame prediction, the encoder needs to find a reference block that matches each image block as much as possible in the previously encoded reference image, so as to minimize the prediction error.

Summary of the invention

Technical issues

Image prediction residual refers to the difference between the image prediction value and the original image. With the continuous advancement of prediction technology, image prediction residual is constantly decreasing, but the residual in the edge area of the image encoding/decoding block is still much larger than the residual in the middle of the encoding/decoding block, and the existing traditional transformation method cannot solve this problem.

In recent years, many position-based transform technologies have emerged, such as PBT (Position based Transform), SBT (Sub-block based Transform), etc. The technical block diagram of PBT technology is shown in Figure 1. PBT is a position-based transform, which is applied to inter-frame prediction residual blocks to better fit the inter-frame residual characteristics. By dividing the current encoding/decoding block into 4 transform blocks of equal size (for example, blocks 0, 1, 2 and 3), and then selecting different transform kernels to transform the transform blocks according to the different positions of different transform blocks in the encoding/decoding block, a more efficient residual signal expression can be achieved. The transform types allowed by PBT include discrete cosine transform 8 (DCT8) and discrete sine transform 7 (DST7). The maximum size of the coding unit using this method is 32x32, the minimum is 8x8, and the aspect ratio of the coding unit is not greater than 2.

The technical block diagram of the SBT technology is shown in Figure 2. The SBT technology assumes the local distribution of inter-frame prediction residuals, that is, the prediction residuals are distributed in the local 1/2 or 1/4 area of the residual block. By dividing the codec block into two transform blocks, and forcing the residual of one transform block to 0, only the other transform block with a non-zero residual is transformed, thereby reducing the high-frequency components of the transform coefficients and improving the compression performance. The above-mentioned PBT technology and SBT technology both effectively remove the residuals at the edge of the codec block and improve the coding efficiency.

However, although both PBT and SBT technologies are transformation methods designed for residual errors at the edge of codec blocks, both PBT and SBT technologies have some limitations. For example, SBT can only remove the redundancy of residual errors at one edge of a codec block; PBT needs to transmit more information in the bitstream to remove the redundancy of residual errors at the edge of the codec block.

Technical Solution

In one aspect of the present invention, a method for transforming in video coding is provided, the method comprising: obtaining a residual of a current coding block according to a prediction value of a current coding block of an image and an original value of the image; dividing the current coding block into n sub-transformation blocks according to mode information for dividing the current coding block into sub-transformation blocks, and performing transforms corresponding to the sub-transformation blocks on the residuals of the generated different sub-transformation blocks, wherein n is an integer greater than or equal to 3; quantizing the result of the transform to obtain coefficients of the sub-transformation blocks, and performing coefficient encoding on the coefficients of the sub-transformation blocks; and writing the result of the coefficient encoding and the mode information for dividing the current coding block into a bitstream.

According to one aspect of the present invention, a method for transforming in video coding is provided, wherein the mode information of dividing the current coding block into sub-transform blocks includes a division direction and information indicating the number of sub-transform blocks into which the current coding block is divided, wherein the division direction is used to indicate that the current coding block is divided in a horizontal direction or a vertical direction.

According to one aspect of the present invention, a method for transformation in video encoding is provided, wherein the residuals of different generated sub-transform blocks are transformed corresponding to the sub-transform blocks, including: when the division direction indicates the horizontal direction and n=3, for the sub-transform block of the uppermost layer, a discrete sine transform 7 (DST7) transform kernel is applied in the horizontal direction, and a discrete cosine transform 8 (DCT8) transform kernel is applied in the vertical direction; for the middle sub-transform blocks, their residuals are forced to be 0 and no transformation is performed; for the sub-transform block of the lowermost layer, a DST7 transform kernel is applied in the horizontal direction, and a DST7 transform kernel is applied in the vertical direction.

According to one aspect of the present invention, a method for transforming in video coding is provided, wherein, when the division direction indicates the horizontal direction and n=3, the residuals of the sub-transform blocks of the uppermost layer and the sub-transform blocks of the lowermost layer are forced to be set to 0, and the sub-transform blocks of the uppermost layer and the sub-transform blocks of the lowermost layer are not transformed, and discrete cosine transform 2 (DCT2) is used as the transform kernel for the intermediate sub-transform blocks in both the horizontal and vertical directions.

According to one aspect of the present invention, a method for transformation in video encoding is provided, wherein the pattern information of dividing the current coding block into sub-transform blocks includes: when the division direction indicates the horizontal direction and n=3, the current coding block is horizontally divided into an uppermost sub-transform block with a height of the current coding block height/4, an intermediate sub-transform block with a height of the current coding block height/2, and a lowermost sub-transform block with a height of the current coding block height/4.

In one aspect of the present invention, a method for transformation in video encoding is provided, wherein the residuals of different generated sub-transform blocks are transformed corresponding to the sub-transform blocks, including: when the division direction indicates the vertical direction and n=3, for the leftmost sub-transform block, the DCT8 transform kernel is applied in the horizontal direction, and the DST7 transform kernel is applied in the vertical direction; for the middle sub-transform block, its residual is forced to be 0, and no transformation is performed; for the rightmost sub-transform block, the DST7 transform kernel is applied in the horizontal direction, and the DST7 transform kernel is applied in the vertical direction.

In one aspect of the present invention, a method for transformation in video coding is provided, wherein the residuals of different generated sub-transform blocks are transformed corresponding to the sub-transform blocks, including: when the division direction indicates the vertical direction and n=3, the residuals of the leftmost sub-transform block and the rightmost sub-transform block are forced to be 0, and the leftmost sub-transform block and the rightmost sub-transform block are not transformed, and DCT2 is used as the transformation kernel in both the horizontal and vertical directions for the middle sub-transform block.

In one aspect of the present invention, a method for transform in video encoding is provided, wherein the mode information of dividing the current coding block into sub-transform blocks includes: when the division direction indicates a vertical direction and n=3, the current coding block is vertically divided into a leftmost sub-transform block with a width of the current coding block width/4, a middle sub-transform block with a width of the current coding block width/2, and a rightmost sub-transform block with a width of the current coding block width/4.

In one aspect of the present invention, a method for transform in video coding is provided, wherein the results of the coefficient encoding of the sub-transform blocks and the mode information of dividing the current coding block into the sub-transform blocks are written into the bitstream, including: writing an identification of whether the current coding block needs to be divided into sub-transform blocks into the bitstream; writing an identification indicating the number of sub-transform blocks into which the current coding block is divided into the bitstream; writing the division direction into the bitstream; and writing the results of the coefficient encoding of the sub-transform blocks into the bitstream.

In one aspect of the present invention, a method for transforming in video encoding is provided, wherein writing an identifier indicating the number of sub-transformation blocks into which a current coding block is divided into a bitstream comprises: when the width and height of the current coding block are both less than a predetermined threshold, not writing the identifier indicating the number of sub-transformation blocks into which the current coding block is divided and the division direction into the bitstream; otherwise, writing the identifier indicating the number of sub-transformation blocks into which the current coding block is divided into the bitstream; and if the number of sub-transformation blocks divided is greater than or equal to 3, writing the division direction into the bitstream.

In one aspect of the present invention, a method for transformation in video encoding is provided, wherein writing a division direction into a bitstream comprises: if a ratio of a width to a height of a current coding block is greater than a predetermined threshold, or a ratio of a width to a height of a current coding block is less than a predetermined threshold, then the division direction is not transmitted; otherwise, the division direction is written into the bitstream.

In one aspect of the present invention, a method for transformation in video encoding is provided, wherein writing a division direction into a bitstream comprises: after writing an identifier of the number of sub-transformation blocks into which a current coding block is divided into the bitstream, if it is determined that the number of sub-transformation blocks divided is greater than or equal to 3, writing the division direction into the bitstream.

According to one aspect of the present invention, a method for transformation in video coding is provided, wherein coefficient encoding of the coefficients of the sub-transform blocks comprises: not writing an identifier CBF indicating whether each sub-transform block contains coefficients, and for the case of n=3, the first sub-transform block coefficient is defaulted to 1, the second sub-transform block coefficient is defaulted to 0, and the third sub-transform block coefficient is defaulted to 1.

According to one aspect of the present invention, a method for transform in video encoding is provided, wherein coefficient encoding of the coefficients of the sub-transform blocks comprises: not writing an identifier CBF indicating whether each sub-transform block contains coefficients, and for the case of n=3, the first sub-transform block coefficient is defaulted to 0, the second sub-transform block coefficient is defaulted to 1, and the third sub-transform block coefficient is defaulted to 0.

In another aspect of the present invention, a method for transform in video decoding is provided, the method comprising: parsing the mode information of dividing the current decoding block into sub-transform blocks and the coefficients of the sub-transform blocks from the bit stream; dividing the current decoding block into n sub-transform blocks according to the mode information of dividing the current coding block into sub-transform blocks, wherein n is an integer greater than or equal to 3; performing inverse transforms corresponding to the sub-transform blocks on the coefficients of different sub-transform blocks to obtain residuals of the sub-transform blocks; and reconstructing the image through the residuals of the sub-transform blocks of the current decoding block and the predicted values of the current decoding block.

Another aspect of the present invention provides a method for transformation in video decoding, wherein the mode information of dividing the current decoding block into sub-transform blocks includes a division direction and block number information indicating that the current decoding block is divided into sub-transform blocks, and the division direction is used to indicate that the current coding block is divided in a horizontal direction or a vertical direction.

Another aspect of the present invention provides a method for transformation in video decoding, wherein parsing the mode information of dividing the current decoding block into sub-transform blocks and the coefficients of the sub-transform blocks from the bitstream includes: parsing an identifier of whether the current decoding block is divided into sub-transform blocks, parsing an identifier indicating the number of sub-transform blocks into which the current decoding block is divided, parsing the division direction of the current decoding block, and parsing the coefficients of the sub-transform blocks.

Another aspect of the present invention provides a method for transformation in video decoding, wherein parsing an identifier indicating the number of sub-transform blocks into which a current decoding block is divided comprises: when the width and height of the current decoding block are both less than a predetermined threshold, not parsing the identifier indicating the number of sub-transform blocks into which the current decoding block is divided and the division direction; otherwise, parsing the identifier indicating the number of sub-transform blocks into which the current decoding block is divided, and if the number of sub-transform blocks is greater than or equal to 3, parsing the division direction.

Another aspect of the present invention provides a method for transformation in video decoding, wherein parsing the division direction includes: if the ratio of the width to the height of the current decoding block is greater than a predetermined threshold, or the ratio of the width to the height of the current decoding block is less than a predetermined threshold, then the division direction is not parsed; otherwise, the division direction is parsed.

Another aspect of the present invention provides a method for transformation in video decoding, wherein parsing the division direction includes: after parsing an identifier indicating the number of sub-transform blocks into which a current decoding block is divided, if it is determined that the number of divided sub-transform blocks is greater than or equal to 3, parsing the division direction.

Another aspect of the present invention provides a method for transformation in video decoding, wherein parsing the coefficients of the sub-transform blocks includes: not parsing the identifier CBF indicating whether each sub-transform block contains coefficients, and for the case of n=3, the coefficient of the first sub-transform block is defaulted to 1, the coefficient of the second sub-transform block is defaulted to 0, and the coefficient of the third sub-transform block is defaulted to 1.

Another aspect of the present invention provides a method for transformation in video decoding, wherein parsing the coefficients of the sub-transform blocks includes: not parsing the identifier CBF indicating whether each sub-transform block contains coefficients, and for the case of n=3, the coefficients of the first sub-transform block are defaulted to 0, the coefficients of the second sub-transform block are defaulted to 1, and the coefficients of the third sub-transform block are defaulted to 0.

On the other hand, the present invention provides a method for transformation in video decoding, wherein, when the division direction indicates the horizontal direction and n=3, for the sub-transform block of the uppermost layer, the DST7 transform kernel is applied in the horizontal direction for inverse transformation, and the DCT8 transform kernel is applied in the vertical direction for inverse transformation; for the middle sub-transform block, its residual is forced to be 0, and no inverse transformation is performed; for the sub-transform block of the lowermost layer, the DST7 transform kernel is applied in the horizontal direction for inverse transformation, and the DST7 transform kernel is applied in the vertical direction for inverse transformation.

Another aspect of the present invention provides a method for transformation in video decoding, wherein the coefficients of different sub-transform blocks are inversely transformed corresponding to the sub-transform blocks, including: when the division direction indicates the horizontal direction and n=3, for the sub-transform block of the uppermost layer, the discrete sine transform 7 (DST7) transform kernel is applied in the horizontal direction for inverse transformation, and the discrete cosine transform 8 (DCT8) transform kernel is applied in the vertical direction for inverse transformation; for the middle sub-transform block, its residual is forced to 0, and no inverse transformation is performed; for the sub-transform block of the lowermost layer, the DST7 transform kernel is applied in the horizontal direction for inverse transformation, and the DST7 transform kernel is applied in the vertical direction for inverse transformation.

Another aspect of the present invention provides a method for transformation in video decoding, wherein the pattern information of dividing the current coding block into sub-transform blocks includes: when the division direction indicates the horizontal direction and n=3, the current coding block is horizontally divided into an uppermost sub-transform block with a height of the current coding block height/4, an intermediate sub-transform block with a height of the current coding block height/2, and a lowermost sub-transform block with a height of the current coding block height/4.

Another aspect of the present invention provides a method for transformation in video decoding, wherein the coefficients of different sub-transform blocks are inversely transformed corresponding to the sub-transform blocks, including: when the division direction indicates the vertical direction and n=3, for the leftmost sub-transform block, the DCT8 transform kernel is applied in the horizontal direction for inverse transformation, and the DST7 transform kernel is applied in the vertical direction for inverse transformation; for the middle sub-transform block, its residual is forced to be 0, and no inverse transformation is performed; for the rightmost sub-transform block, the DST7 transform kernel is applied in the horizontal direction for inverse transformation, and the DST7 transform kernel is applied in the vertical direction for inverse transformation.

Another aspect of the present invention provides a method for transformation in video decoding, wherein performing inverse transformation on coefficients of different sub-transform blocks corresponding to the sub-transform blocks comprises: when the division direction indicates the vertical direction and n=3, forcing the residuals of the leftmost sub-transform block and the rightmost sub-transform block to be 0, and not performing inverse transformation on the leftmost sub-transform block and the rightmost sub-transform block, while performing inverse transformation on the middle sub-transform block in both the horizontal and vertical directions using DCT2 as the transformation kernel.

In another aspect of the present invention, a method for transformation in video decoding is provided, wherein the mode information of dividing the current coding block into sub-transform blocks includes: when the division direction indicates a vertical direction and n=3, the current coding block is vertically divided into a leftmost sub-transform block with a width of the current coding block width/4, a middle sub-transform block with a width of the current coding block width/2, and a rightmost sub-transform block with a width of the current coding block width/4.

In another aspect of the present invention, a device for performing a method for transforming in video image encoding is provided, the device comprising: a prediction unit, which is configured to receive image data and is used to predict a predicted value of a current coding block in the image data, and obtain a residual of the current coding block according to the original value of the image data; a transform unit, which is coupled to the prediction unit and is configured to divide the current coding block of the image data into n sub-transform blocks according to mode information of dividing the current coding block into sub-transform blocks, and perform a transform corresponding to the sub-transform block on the residuals of the generated different sub-transform blocks, and obtain coefficients of the sub-transform blocks by quantizing the transform results, and perform coefficient encoding on the coefficients of the sub-transform blocks, wherein n is an integer greater than or equal to 3; and a writing unit, which is coupled to the conversion unit and is configured to write the mode information of dividing the current coding block into sub-transform blocks and the result of the coefficient encoding into a bitstream.

According to another aspect of the present invention, a device for performing a method for transforming in coding is provided, wherein the mode information of dividing the current coding block into sub-transform blocks includes a division direction and information indicating the number of blocks into which the current coding block is divided, wherein the division direction is used to indicate that the current coding block is divided in a horizontal direction or a vertical direction.

On the other hand, the present invention provides an apparatus for performing a method for transforming in encoding, wherein the transform unit is configured to perform a transform corresponding to the sub-transform block on the residuals of different generated sub-transform blocks, including: when the division direction indicates the horizontal direction and n=3, for the sub-transform block of the uppermost layer, a discrete sine transform 7 (DST7) transform kernel is applied in the horizontal direction, and a discrete cosine transform 8 (DCT8) transform kernel is applied in the vertical direction; for the middle sub-transform block, its residual is forced to be 0 and no transformation is performed; for the sub-transform block of the lowermost layer, a DST7 transform kernel is applied in the horizontal direction, and a DST7 transform kernel is applied in the vertical direction.

In another aspect of the present invention, a device for performing a method for transforming in encoding is provided, wherein the pattern information of dividing the current coding block into sub-transform blocks includes: when the division direction indicates a horizontal direction and n=3, the current coding block is horizontally divided into an uppermost sub-transform block with a height of the current coding block height/4, an intermediate sub-transform block with a height of the current coding block height/2, and a lowermost sub-transform block with a height of the current coding block height/4.

On the other hand, the present invention provides an apparatus for performing a method for transforming in encoding, wherein the transform unit is configured to perform a transform corresponding to the sub-transform block on the residuals of different generated sub-transform blocks, including: when the division direction indicates a vertical direction and n=3, for the leftmost sub-transform block, applying the DCT8 transform kernel in the horizontal direction and the DST7 transform kernel in the vertical direction; for the middle sub-transform block, forcing its residual to 0 and not performing the transform; for the rightmost sub-transform block, applying the DST7 transform kernel in the horizontal direction and the DST7 transform kernel in the vertical direction.

In another aspect of the present invention, a device for performing a method for transforming in coding is provided, wherein the mode information of dividing the current coding block into sub-transform blocks includes: when the division direction indicates a vertical direction and n=3, the current coding block is vertically divided into a leftmost sub-transform block with a width of the current coding block width/4, a middle sub-transform block with a width of the current coding block width/2, and a rightmost sub-transform block with a width of the current coding block width/4.

In another aspect of the present invention, a device for performing a method for transforming in coding is provided, wherein the writing unit is configured to write mode information of dividing a current coding block into sub-transform blocks and the result of the coefficient encoding into a bitstream, including: writing an identification of whether the current coding block needs to be divided into sub-transform blocks into the bitstream; writing an identification indicating the number of sub-transform blocks into which the current coding block is divided into the bitstream; writing a division direction into the bitstream; and writing the result of the coefficient encoding of the sub-transform block into the bitstream.

In another aspect of the present invention, a device for performing a method for transforming in coding is provided, wherein the writing unit is configured to write an identifier indicating the number of sub-transformation blocks into which a current coding block is divided into into a bitstream, including: when the width and height of the current coding block are both less than a predetermined threshold, the identifier indicating the number of sub-transformation blocks into which the current coding block is divided and the division direction are not written into the bitstream; otherwise, the identifier indicating the number of sub-transformation blocks into which the current coding block is divided is written into the bitstream; if the number of sub-transformation blocks divided is greater than or equal to 3, the division direction is written into the bitstream.

Another aspect of the present invention provides an apparatus for performing a method for transforming in encoding, wherein the writing unit is configured to write a division direction into a bitstream, including: if the ratio of the width to the height of the current coding block is greater than a predetermined threshold, or the ratio of the width to the height of the current coding block is less than a predetermined threshold, then the division direction is not transmitted; otherwise, the division direction is written into the bitstream.

Another aspect of the present invention provides an apparatus for performing a method for transforming in coding, wherein the transform unit is configured to perform coefficient encoding on the coefficients of the sub-transform block, including: not writing an identifier CBF indicating whether each sub-transform block contains a coefficient, and for the case of n=3, the first sub-transform block coefficient is defaulted to 1, the second sub-transform block coefficient is defaulted to 0, and the third sub-transform block coefficient is defaulted to 1.

In another aspect of the present invention, a device for performing a method for transforming in video image decoding is provided, the device comprising: a parsing unit, which is configured to receive a bit stream and to parse the mode information of dividing a current decoding block into sub-transform blocks and the coefficients of the sub-transform blocks; a transform unit, which is coupled to the parsing unit and is configured to divide the current decoding block into n sub-transform blocks according to the mode information of dividing the sub-transform blocks of the current coding block, and perform an inverse transform corresponding to the sub-transform blocks on the coefficients of different sub-transform blocks to obtain residuals of the sub-transform blocks, wherein n is an integer greater than or equal to 3; and a reconstruction unit, which is coupled to the conversion unit and is configured to reconstruct an image through the residuals of the sub-transform blocks of the current decoding block and the predicted value of the current decoding block.

In another aspect of the present invention, a device is provided for executing a method for performing transformation in video image decoding, wherein the mode information of dividing the current decoding block into sub-transformation blocks includes a division direction and block number information indicating that the current decoding block is divided into sub-transformation blocks, and the division direction is used to indicate that the current coding block is divided in a horizontal direction or a vertical direction.

In another aspect of the present invention, a device for executing a method for performing transformation in video image decoding is provided, wherein the parsing unit is configured to parse the mode information of dividing the current decoding block into sub-transform blocks and the coefficients of the sub-transform blocks from the bitstream, including: parsing an identifier of whether the current decoding block is divided into sub-transform blocks, parsing an identifier indicating the number of sub-transform blocks into which the current decoding block is divided, parsing the division direction of the current decoding block, and parsing the coefficients of the sub-transform blocks.

In another aspect of the present invention, a device for executing a method for performing transformation in video image decoding is provided, wherein the parsing unit is configured to parse an identifier indicating the number of sub-transformation blocks into which a current decoding block is divided, including: when the width and height of the current decoding block are both less than a predetermined threshold, not parsing the identifier indicating the number of sub-transformation blocks into which the current decoding block is divided and the division direction; otherwise, parsing the identifier indicating the number of sub-transformation blocks into which the current decoding block is divided, and if the number of sub-transformation blocks is greater than or equal to 3, parsing the division direction.

Another aspect of the present invention provides an apparatus for executing a method for transformation in video image decoding, wherein the parsing unit is configured to parse the division direction, including: if the ratio of the width to the height of the current decoded block is greater than a predetermined threshold, or the ratio of the width to the height of the current decoded block is less than the predetermined threshold, then the division direction is not parsed; otherwise, the division direction is parsed.

In another aspect of the present invention, a device for executing a method for transformation in video image decoding is provided, wherein the parsing unit is configured to parse coefficients of sub-transform blocks, including: not parsing an identifier CBF indicating whether each sub-transform block contains coefficients, and for the case of n=3, the coefficient of the first sub-transform block is defaulted to 1, the coefficient of the second sub-transform block is defaulted to 0, and the coefficient of the third sub-transform block is defaulted to 1.

On the other hand, the present invention provides an apparatus for executing a method for transformation in video image decoding, wherein the transformation unit is configured to perform inverse transformation on coefficients of different sub-transform blocks corresponding to the sub-transform blocks, including: when the division direction indicates the horizontal direction and n=3, for the sub-transform block of the uppermost layer, applying the discrete sine transform 7 (DST7) transform kernel in the horizontal direction for inverse transformation, and applying the discrete cosine transform 8 (DCT8) transform kernel in the vertical direction for inverse transformation; for the middle sub-transform block, its residual is forced to 0, and no inverse transformation is performed; for the sub-transform block of the lowermost layer, applying the DST7 transform kernel in the horizontal direction for inverse transformation, and applying the DST7 transform kernel in the vertical direction for inverse transformation.

In another aspect of the present invention, a device for performing a method for transformation in video image decoding is provided, wherein the pattern information of dividing the current coding block into sub-transform blocks includes: when the division direction indicates the horizontal direction and n=3, the current coding block is horizontally divided into an uppermost sub-transform block with a height of the current coding block height/4, an intermediate sub-transform block with a height of the current coding block height/2, and a lowermost sub-transform block with a height of the current coding block height/4.

On the other hand, the present invention provides an apparatus for executing a method for transformation in video image decoding, wherein the transformation unit is configured to perform inverse transformation on coefficients of different sub-transform blocks corresponding to the sub-transform blocks, including: when the division direction indicates a vertical direction and n=3, for the leftmost sub-transform block, applying the DCT8 transform kernel in the horizontal direction for inverse transformation, and applying the DST7 transform kernel in the vertical direction for inverse transformation; for the middle sub-transform block, its residual is forced to be 0, and no inverse transformation is performed; for the rightmost sub-transform block, applying the DST7 transform kernel in the horizontal direction for inverse transformation, and applying the DST7 transform kernel in the vertical direction for inverse transformation.

In another aspect of the present invention, a device for performing a method for transforming in video image decoding is provided, wherein the mode information of dividing the current coding block into sub-transform blocks includes: when the division direction indicates a vertical direction and n=3, the current coding block is vertically divided into a leftmost sub-transform block with a width of the current coding block width/4, a middle sub-transform block with a width of the current coding block width/2, and a rightmost sub-transform block with a width of the current coding block width/4.

Other technical features may be apparent to those skilled in the art from the following drawings, descriptions, and claims.

Advantageous Effects of the Invention

The present invention divides the current encoding/decoding block into multiple transform units according to the residual characteristics of the encoding/decoding block, as shown in Figure 3, and designs transforms corresponding to the transform blocks at different positions. The residuals of two edges can be processed simultaneously, effectively removing the redundancy of the residuals at the edge positions of the encoding/decoding block, while reducing the extra overhead in the code stream, thereby improving the coding performance.

BRIEF DESCRIPTION OF THE DRAWINGS

For a more complete understanding of the present disclosure and its advantages, reference is now made to the following description in conjunction with the accompanying drawings, wherein like reference numerals represent like parts:

FIG1 shows a technical block diagram of the PBT technology;

FIG2 shows a technical block diagram of the SBT technology;

FIG3 is a schematic diagram showing a current encoding/decoding block divided into a plurality of sub-transform blocks according to an embodiment of the present disclosure;

FIG4 shows a flow chart of a method for transforming in video image encoding according to an embodiment of the present disclosure;

FIG5 shows a flow chart of a method for transforming in video image decoding according to an embodiment of the present disclosure;

FIG6 shows a block diagram of a device for performing a method for transforming in video image coding according to an embodiment of the present disclosure; and

FIG. 7 shows a block diagram of a device for performing a method for transforming in video image decoding according to an embodiment of the present disclosure.

DETAILED DESCRIPTION

Before proceeding to the following detailed description, it may be advantageous to set forth the definitions of certain words and phrases used throughout this patent document. The term "coupling" and its derivatives refer to any direct or indirect communication between two or more elements, whether or not those elements are in physical contact with each other. The terms "transmission", "receiving" and "communication" and their derivatives cover direct and indirect communication. The terms "include" and "comprising" and their derivatives refer to including but not limited to. The term "or" is inclusive, meaning and/or. The phrase "associated with..." and its derivatives refer to including, including within, interconnecting, containing, contained within, connecting or connecting with, coupling or coupling with, communicating with, cooperating, interweaving, parallel, approaching, binding or binding with, having, having attributes, having a relationship or having a relationship with, etc. The term "controller" refers to any device, system or part thereof that controls at least one operation. Such a controller can be implemented with hardware, or a combination of hardware and software and/or firmware. The functions associated with any particular controller can be centralized or distributed, whether local or remote. The phrase "at least one of", when used with a list of items, means that different combinations of one or more of the listed items may be used, and only one of the items in the list may be required. For example, "at least one of A, B, C" includes any of the following combinations: A, B, C, A and B, A and C, B and C, A and B and C.

Definitions for other specific words and phrases are provided throughout this patent document. Those of ordinary skill in the art should understand that in many, if not most instances, such definitions apply to prior and future uses of such defined words and phrases.

FIG. 1 to FIG. 7 discussed below and the various embodiments used to describe the principles of the present disclosure in this patent document are merely exemplary and should not be interpreted in any way to limit the scope of the present disclosure. Those skilled in the art will appreciate that the principles of the present disclosure can be implemented in any suitably arranged system or device.

FIG3 shows a schematic diagram of dividing a current encoding/decoding block into a plurality of sub-transform blocks according to an embodiment of the present disclosure. The embodiment shown in FIG3 is for illustration only. Other division methods may be used without departing from the scope of the present disclosure.

As shown in FIG3(a), the current encoding/decoding block is horizontally divided into three parts (sub-transform blocks), among which: the top layer is the residual part (301), for which the DST7 transform kernel is applied in the horizontal direction and the DCT8 transform kernel is applied in the vertical direction; the middle layer is the non-residual part (302), whose residual is forced to be set to 0; the bottom layer is the residual part (303), for which the DST7 transform kernel is applied in the horizontal direction and the DST7 transform kernel is applied in the vertical direction. In addition, in this embodiment, the heights of the three sub-transform blocks are 1/4, 1/2 and 1/4 of the height of the current encoding/decoding block, respectively. The above-mentioned application combination of transform kernels and the division height of sub-transform blocks are only for illustration. Without departing from the scope of the present disclosure, the application combination of transform kernels and the division height of sub-transform blocks can have different modes.

As shown in FIG3( b ), the current encoding/decoding block is vertically divided into three parts (sub-transform blocks), among which: the leftmost part is a residual part (311), for which the DCT8 transform kernel is applied in the horizontal direction and the DST7 transform kernel is applied in the vertical direction; the middle part is a non-residual part (312), whose residual is forced to be set to 0; the rightmost part is a residual part (313), for which the DST7 transform kernel is applied in the horizontal direction and the DST7 transform kernel is applied in the vertical direction. In addition, in this embodiment, the widths of the three sub-transform blocks are 1/4, 1/2 and 1/4 of the width of the current encoding/decoding block, respectively. The above-mentioned application combination of transform kernels and the division height of sub-transform blocks are only for illustration. Without departing from the scope of the present disclosure, the application combination of transform kernels and the division height of sub-transform blocks can have different modes.

Figure 4 shows a flow chart of a method for transforming in coding according to an embodiment of the present disclosure. The embodiment of the method for transforming in coding shown in Figure 4 is for illustration only, and the embodiment shown in Figure 4 may be adjusted and modified without departing from the scope of the present disclosure.

As shown in FIG4 , the method for transforming in coding starts at step 101. In step 101, the residual of the current coding block is obtained according to the prediction value of the current coding block and the original value of the image. For the current coding block, the encoder needs to find a reference block with a high matching degree with the current coding block in the reference image that has been encoded before the current coding block, and generate the prediction value of the current coding block based on the reference block. Without departing from the scope of the present disclosure, various methods for generating prediction values in the art can be used to generate the prediction value of the current coding block. By comparing the prediction value of the previous coding block with the original value of the image, the residual of the current coding block is obtained.

In step 102, the current coding block is divided into n sub-transform blocks, and the residuals of different generated sub-transform blocks are transformed corresponding to the sub-transform blocks; the coefficients of the sub-transform blocks are obtained by quantizing the transformation results, and the coefficients of the sub-transform blocks are coefficient encoded, where n is an integer greater than or equal to 3.

There are many ways to divide the current coding block: in the division direction, the coding block can be divided in the horizontal direction, as shown in Figure 3(a); the coding block can also be divided in the vertical direction, as shown in Figure 3(b).

In addition, there are many ways to divide the sub-transform block size. One possible way is as shown in Figure 3(a), where the current coding block (width is width, height is height) is horizontally divided into three sub-transform blocks, and the width and height of the three sub-transform blocks are (width, height/4), (width, height/2) and (width, height/4) respectively; it is also possible to divide the current coding block (width is width, height is height) vertically into three sub-transform blocks, as shown in Figure 3(b), where the width and height of the three sub-transform blocks are (width/4, height), (width/2, height) and (width/4, height) respectively.

There may be different ways to perform a transform corresponding to the sub-transform block on the generated residual of the different sub-transform blocks.

One possible approach is shown in FIG3(a): for the uppermost sub-transform block (301), the DST7 transform kernel is applied in the horizontal direction, and the DCT8 transform kernel is applied in the vertical direction; for the middle sub-transform block (302), its residual is forced to be 0, so no transformation is required; for the lowermost sub-transform block (303), the DST7 transform kernel is applied in the horizontal direction, and the DST7 transform kernel is applied in the vertical direction.

The above application method is only used as an example. For the division situation of Figure 3(a), different transformation kernel application methods can also be used. For example, the residuals of the uppermost sub-transform block (301) and the lowermost sub-transform block (303) can be forced to be 0. Therefore, the two sub-transform blocks do not need to be transformed, and DCT2 is used as the transformation kernel in both the horizontal and vertical directions for the middle sub-transform block (302).

Another possible way to perform a transform corresponding to the generated residuals of different sub-transform blocks is shown in Figure 3(b): for the leftmost sub-transform block (311), the DCT8 transform kernel is applied in the horizontal direction and the DST7 transform kernel is applied in the vertical direction; for the middle sub-transform block (312), its residual is forced to 0, so no transformation is required; for the rightmost sub-transform block (313), the DST7 transform kernel is applied in the horizontal direction and the DST7 transform kernel is applied in the vertical direction.

The above application method is only used as an example. For the division situation of Figure 3(b), different transformation kernel application methods can also be used. For example, the residuals of the leftmost sub-transform block (311) and the rightmost sub-transform block (313) can be forced to be 0. Therefore, the two sub-transform blocks do not need to be transformed, and DCT2 is used as the transformation kernel in both the horizontal and vertical directions for the middle sub-transform block (312).

In step 103, the result of the coefficient encoding and the corresponding mode of dividing the encoding block into sub-transform blocks are written into the bitstream.

Different methods can be used to write the coefficient encoding result and the corresponding mode of dividing the coding block into sub-transform blocks into the bitstream. One possible implementation method is: first write the identification of whether the current coding block needs to be divided into sub-transform blocks into the bitstream, then write the identification indicating the number of blocks into which the current coding block is divided into sub-transform blocks into the bitstream, then write the division direction into the bitstream, and finally write the result of the coefficient encoding of the sub-transform block into the bitstream. The order of the above writing can also be appropriately changed and adjusted within the scope of the present invention.

A possible method for writing the identifier indicating the number of sub-transformation blocks into which the current coding block is divided into a bitstream is as follows: when the width and height of the current coding block are both smaller than a predetermined threshold, the identifier indicating the number of sub-transformation blocks into which the current coding block is divided and the division direction are not written into the bitstream; otherwise, the identifier indicating the number of sub-transformation blocks into which the current coding block is divided is written into the bitstream; if the number of sub-transformation blocks divided is greater than or equal to 3, the division direction continues to be written into the bitstream.

Among them, a possible method of writing the division direction into the bitstream is: if the ratio of the width to the height of the current coding block is greater than a predetermined threshold, or the ratio of the width to the height of the current coding block is less than a predetermined threshold, then the division direction does not need to be transmitted, otherwise the division direction is written into the bitstream.

Another possible method of writing the division direction into the bitstream is: after writing the identifier of the number of sub-transformation blocks into which the current coding block is divided into the bitstream, if it is determined that the number of sub-transformation blocks divided is greater than or equal to 3, the division direction is written into the bitstream.

Among them, a possible method of writing the coefficient encoding result of the coefficient of the sub-transform block into the bitstream is: do not write an identifier (CBF, coding block flag) indicating whether each sub-transform block contains a coefficient. For the case where the current coding block is divided into 3 sub-transform blocks, the encoding end defaults to the first sub-transform block coefficient being 1, the second sub-transform block coefficient being 0, and the third sub-transform block coefficient being 1.

In addition, a possible method for writing the coefficient encoding results of the coefficients of the sub-transform block into the bitstream is: do not write an identifier (CBF, coding block flag) indicating whether each sub-transform block contains a coefficient. For the case where the current coding block is divided into 3 sub-transform blocks, the encoding end defaults the coefficient of the first sub-transform block to 0, the coefficient of the second sub-transform block to 1, and the coefficient of the third sub-transform block to 0.

Figure 5 shows a flow chart of a method for transforming in decoding according to an embodiment of the present disclosure. The embodiment of the method for transforming in decoding shown in Figure 5 is for illustration only, and adjustments and modifications may be made to the embodiment shown in Figure 5 without departing from the scope of the present disclosure.

As shown in Fig. 5, the method for transform in decoding starts at step 201. At step 201, the mode information of dividing the current decoding block into sub-transform blocks and the coefficients of the sub-transform blocks (residuals of the sub-transform blocks) are parsed from the code stream.

There are multiple methods for parsing the mode of dividing the current decoded block into sub-transform blocks and the coefficients of the sub-transform blocks from the bitstream. One possible implementation method is: first parse the flag of whether the current decoded block is divided into sub-transform blocks, then parse the flag indicating the number of sub-transform blocks into which the current decoded block is divided, then parse the division direction of the current decoded block, and finally parse the coefficients of the sub-transform blocks. The above parsing order can also be appropriately changed and adjusted within the scope of the present invention.

A possible method for parsing the identifier indicating the number of sub-transform blocks into which the current decoding block is divided is as follows: when the width and height of the current decoding block are both smaller than a predetermined threshold, the identifier indicating the number of sub-transform blocks into which the current decoding block is divided and the division direction are not parsed, and the technology of the present application is not adopted at this time; otherwise, the identifier indicating the number of sub-transform blocks into which the current decoding block is divided is parsed, and if the number of sub-transform blocks is greater than or equal to 3, the division direction is continued to be parsed.

A possible method for parsing the division direction is: if the ratio of the width to the height of the current decoded block is greater than a predetermined threshold, or the ratio of the width to the height of the current decoded block is less than a predetermined threshold, then the division direction is not parsed; otherwise, the division direction is parsed.

Another possible method for parsing the division direction is: after parsing the number of blocks of the sub-transform block, if the number of blocks of the divided sub-transform block is greater than or equal to 3, then the division direction is parsed.

Among them, a possible method for parsing the coefficients of the sub-transform blocks is: not parsing the flag (CBF, coding block flag) indicating whether each sub-transform block contains coefficients. When the current decoding block is divided into 3 sub-transform blocks, the decoding end defaults to the first sub-transform block coefficient being 1, the second sub-transform block coefficient being 0, and the third sub-transform block coefficient being 1.

In addition, a possible method for parsing the coefficients of the sub-transform blocks is: not parsing the flag (CBF, coding block flag) indicating whether each sub-transform block contains coefficients. When the current decoding block is divided into 3 sub-transform blocks, the decoding end defaults to the first sub-transform block coefficient being 0, the second sub-transform block coefficient being 1, and the third sub-transform block coefficient being 0.

In step 202, the current decoding block is divided into n sub-transform blocks according to the corresponding mode of dividing the coding block into sub-transform blocks, and the coefficients of different sub-transform blocks are inversely transformed corresponding to the sub-transform blocks to obtain residuals of the sub-transform blocks, where n is an integer greater than or equal to 3.

There are many ways to divide the current decoding block: in the division direction, the decoding block can be divided in the horizontal direction, as shown in FIG. 3( a ); the decoding block can also be divided in the vertical direction, as shown in FIG. 3( b ).

In addition, there are many ways to divide the sub-transform block size. One possible way is shown in Figure 3(a), where the current decoding block (width is width, height is height) is horizontally divided into three sub-transform blocks, and the width and height of the three sub-transform blocks are (width, height/4), (width, height/2) and (width, height/4) respectively; it is also possible to divide the current decoding block (width is width, height is height) vertically into three sub-transform blocks, as shown in Figure 3(b), where the width and height of the three sub-transform blocks are (width/4, height), (width/2, height) and (width/4, height) respectively.

There may be different ways to perform inverse transformation on the generated different sub-transform blocks using the transformation kernel corresponding to the sub-transform blocks.

One possible approach is shown in FIG3(a): for the uppermost sub-transform block (301), the DST7 transform kernel is applied in the horizontal direction for inverse transform, and the DCT8 transform kernel is applied in the vertical direction for inverse transform; for the middle sub-transform block (302), its residual is forced to be set to 0, so there is no need for inverse transform; for the lowermost sub-transform block (303), the DST7 transform kernel is applied in the horizontal direction for inverse transform, and the DST7 transform kernel is applied in the vertical direction for inverse transform.

The above application method is only used as an example. For the division situation of Figure 3(a), different transformation kernel application methods can also be used. For example, the residuals of the uppermost sub-transform block (301) and the lowermost sub-transform block (303) can be forced to be 0. Therefore, there is no need to perform inverse transformation on the two sub-transform blocks, and DCT2 is used as the transformation kernel for inverse transformation in both the horizontal and vertical directions for the middle sub-transform block (302).

Another possible way to perform inverse transforms corresponding to the generated different sub-transform blocks is shown in Figure 3(b): for the leftmost sub-transform block (311), the DCT8 transform kernel is applied in the horizontal direction for inverse transform, and the DST7 transform kernel is applied in the vertical direction for inverse transform; for the middle sub-transform block (312), its residual is forced to 0, so no inverse transform is required; for the rightmost sub-transform block (313), the DST7 transform kernel is applied in the horizontal direction for inverse transform, and the DST7 transform kernel is applied in the vertical direction for inverse transform.

The above application method is only used as an example. For the division situation of Figure 3(b), different transformation kernel application methods can also be used. For example, the residuals of the leftmost sub-transform block (311) and the rightmost sub-transform block (313) can be forced to be 0. Therefore, there is no need to perform inverse transformation on these two sub-transform blocks, and DCT2 is used as the transformation kernel for inverse transformation in both the horizontal and vertical directions for the middle sub-transform block (312).

At step 203, the image is reconstructed by the residual of the sub-transform block of the current decoded block. The current decoded block is reconstructed by the residual of the sub-transform block obtained in step 202 and the prediction value of the current decoded block, thereby achieving image reconstruction.

FIG6 shows a block diagram of an apparatus for performing a method for transforming in coding according to an embodiment of the present disclosure. The embodiment of the apparatus for performing a method for transforming in coding shown in FIG6 is for illustration only, and adjustments and modifications may be made to the embodiment shown in FIG6 without departing from the scope of the present disclosure.

In FIG6 , the apparatus for performing the method of transform in coding includes: a prediction unit (601), a transform unit (602) and a write unit (603). The prediction unit (601) receives original image data and is used to predict the prediction value of the current coding block in the image data, and obtain the residual of the current coding block according to the original value of the image data. The transform unit (602) is coupled to the prediction unit (601) and is configured to divide the current coding block of the image data into n sub-transform blocks according to the mode information of dividing the current coding block into sub-transform blocks, and perform a transform corresponding to the sub-transform block on the residuals of the generated different sub-transform blocks; obtain the coefficients of the sub-transform blocks by quantizing the transform results, and perform coefficient encoding on the coefficients of the sub-transform blocks, wherein n is an integer greater than or equal to 3. The write unit (603) is coupled to the conversion unit (602) and writes the corresponding mode of dividing the coding block into the sub-transform blocks and the coefficient encoding result of the sub-transform blocks into the bitstream.

FIG7 shows a block diagram of an apparatus for performing a method for transforming in decoding according to an embodiment of the present disclosure. The embodiment of the apparatus for performing a method for transforming in decoding shown in FIG7 is for illustration only, and adjustments and modifications may be made to the embodiment shown in FIG7 without departing from the scope of the present disclosure.

Referring to FIG7 , the apparatus for performing the method of transform in decoding includes: a parsing unit (701), a transform unit (702) and a reconstruction unit (703). The parsing unit (701) receives a code stream from an encoding end and is used to parse the mode information of dividing the current decoding block into sub-transform blocks and the coefficients of the sub-transform blocks. The transform unit (702) is coupled to the parsing unit (701) and is configured to divide the current decoding block into n sub-transform blocks according to the corresponding mode of dividing the coding block into sub-transform blocks, and perform an inverse transform corresponding to the sub-transform blocks on the coefficients of different sub-transform blocks to obtain the residual of the sub-transform block, wherein n is an integer greater than or equal to 3. The reconstruction unit (703) is coupled to the conversion unit (702) and is configured to reconstruct an image through the residual of the sub-transform block of the current decoding block and the prediction value of the current decoding block.

Although the present disclosure has been described with exemplary embodiments, various changes and modifications may be suggested to one skilled in the art. The present disclosure is intended to encompass such changes and modifications as fall within the scope of the appended claims.

Any description in this application should not be construed as implying that any particular element, step, or function is an essential element that must be included in the claims scope. The scope of the patented subject matter is limited solely by the claims.

Claims (50)

1. A method for transforming in video image encoding, the method comprising:

obtaining a residual error of the current coding block according to the predicted value of the current coding block of the image and the original value of the image;

Dividing the current coding block into n sub-transformation blocks according to mode information for dividing the current coding block into the sub-transformation blocks, and transforming residuals of the generated different sub-transformation blocks corresponding to the sub-transformation blocks, wherein n is an integer greater than or equal to 3;

obtaining coefficients of a sub-transform block by quantizing a result of the transform, and coefficient-encoding the coefficients of the sub-transform block; and

Writing the result of the coefficient coding and the mode information of the current coding block dividing sub-transform blocks into code streams,

Wherein, when the division direction indicates the horizontal direction, residual errors of sub-transform blocks of a specific layer of the horizontal division are forcibly set to 0 and the sub-transform blocks thereof are not transformed, and corresponding transform kernels including a discrete sine transform 7 (DST 7) transform kernel, a discrete cosine transform 8 (DCT 8) transform kernel, and a discrete cosine transform 2 (DCT 2) are applied to the sub-transform blocks of other layers than the specific layer; or alternatively

Wherein when the division direction indicates the vertical direction, residuals of sub-transform blocks of a specific portion of the vertical division are forcibly set to 0 and are not transformed, and corresponding transform kernels including a discrete sine transform 7 (DST 7) transform kernel, a discrete cosine transform 8 (DCT 8) transform kernel, and a discrete cosine transform 2 (DCT 2) are applied to the sub-transform blocks of other portions except the specific portion.

2. The method of claim 1, wherein the mode information of dividing the current encoding block into sub-transform blocks includes a division direction indicating division of the current encoding block in a horizontal direction or a vertical direction and block number information indicating division of the current encoding block into sub-transform blocks.

3. The method of claim 2, wherein transforming the residuals of the generated different sub-transform blocks corresponding to the sub-transform block comprises: when the division direction indicates a horizontal direction and n=3, for the uppermost sub-transform block, a discrete sine transform 7 (DST 7) transform kernel is applied in the horizontal direction and a discrete cosine transform 8 (DCT 8) transform kernel is applied in the vertical direction; for the middle sub-transform block, its residual is forced to 0 and no transform is performed; for the sub transform block of the lowest layer, DST7 transform kernels are applied in the horizontal direction, and DST7 transform kernels are applied in the vertical direction.

4. The method of claim 2, wherein when the division direction indicates a horizontal direction and n=3, residuals of an uppermost sub-transform block and a lowermost sub-transform block are forcedly set to 0, and the uppermost sub-transform block and the lowermost sub-transform block are not transformed, and a discrete cosine transform 2 (DCT 2) is employed as a transform kernel for an intermediate sub-transform block in both horizontal and vertical directions.

5. The method of claim 2, wherein the mode information of the current coding block partition sub-transform block comprises: when the division direction indicates a horizontal direction and n=3, the current coding block is horizontally divided into an uppermost sub-transform block having a height of current coding block height/4, an intermediate sub-transform block having a height of current coding block height/2, and a lowermost sub-transform block having a height of current coding block height/4.

6. The method of claim 2, wherein transforming the residuals of the generated different sub-transform blocks corresponding to the sub-transform block comprises: when the division direction indicates a vertical direction and n=3, for the leftmost sub transform block, a DCT8 transform kernel is applied in a horizontal direction and a DST7 transform kernel is applied in a vertical direction; forcing its residual to 0 for the middle sub-transform block and not transforming; for the rightmost sub-transform block, the DST7 transform kernel is applied in the horizontal direction and the DST7 transform kernel is applied in the vertical direction.

7. The method of claim 2, wherein transforming the residuals of the generated different sub-transform blocks corresponding to the sub-transform block comprises: when the division direction indicates a vertical direction and n=3, residuals of the leftmost sub-transform block and the rightmost sub-transform block are forcibly set to 0, and the leftmost sub-transform block and the rightmost sub-transform block are not transformed, and DCT2 is employed as a transform kernel for the middle sub-transform block in both the horizontal direction and the vertical direction.

8. The method of claim 2, wherein the mode information of the current coding block partition sub-transform block comprises: when the division direction indicates a vertical direction and n=3, the current encoding block is vertically divided into a leftmost sub-transform block having a width of current encoding block width/4, an intermediate sub-transform block having a width of current encoding block width/2, and a rightmost sub-transform block having a width of current encoding block width/4.

9. The method of claim 1, writing the result of coefficient encoding of the sub-transform block and mode information of the current encoded block divided sub-transform block into a bitstream comprises:

Writing the identification of whether the current coding block needs to be divided into sub-transformation blocks into a code stream;

writing an identification indicating the number of blocks of the current encoded block divided into sub-transform blocks into a code stream;

Writing the dividing direction into the code stream; and

And writing the result of coefficient coding of the sub-transformation block into the code stream.

10. The method of claim 9, writing an identification indicating the number of blocks of the current encoded block divided into sub-transform blocks into the code stream comprising:

When the width and the height of the current coding block are smaller than the preset threshold value, the code stream is not written with the identification indicating the number of blocks of the current coding block divided into sub-transformation blocks and the dividing direction, otherwise, the code stream is written with the identification indicating the number of blocks of the current coding block divided into sub-transformation blocks, and if the number of blocks of the divided sub-transformation blocks is larger than or equal to 3, the code stream is written with the dividing direction.

11. The method of claim 9, writing the partitioning direction into the code stream comprises:

if the ratio of the width to the height of the current coding block is greater than a predetermined threshold value or the ratio of the width to the height of the current coding block is less than the predetermined threshold value, the dividing direction is not transmitted, otherwise the dividing direction is written into the code stream.

12. The method of claim 9, writing the partitioning direction into the code stream comprises:

After the identification of the number of blocks of the current coding block divided into sub-transform blocks is written into the code stream, if it is determined that the number of blocks of the divided sub-transform blocks is 3 or more, the division direction is written into the code stream.

13. The method of claim 1, coefficient encoding coefficients of the sub-transform block comprising:

the flag CBF indicating whether or not a coefficient is contained in each sub-transform block is not written, and for the case where n=3, the first sub-transform block coefficient is defaulted to 1, the second sub-transform block coefficient is 0, and the third sub-transform block coefficient is 1.

14. The method of claim 1, coefficient encoding coefficients of the sub-transform block comprising:

The flag CBF indicating whether or not a coefficient is contained in each sub-transform block is not written, and for the case where n=3, the first sub-transform block coefficient is defaulted to 0, the second sub-transform block coefficient is 1, and the third sub-transform block coefficient is 0.

15. A method for transforming in video image decoding, the method comprising:

Analyzing the mode information of the current decoding block dividing sub-transformation block and the coefficient of the sub-transformation block from the code stream;

dividing a current decoding block into n sub-transform blocks according to mode information of the sub-transform blocks divided by the current encoding block, wherein n is an integer greater than or equal to 3;

Performing inverse transformation corresponding to the sub-transformation block on coefficients of different sub-transformation blocks to obtain residual errors of the sub-transformation block; and

An image is reconstructed by the residual of the sub-transformed block of the current decoded block and the prediction value of the current decoded block,

Wherein, when the division direction indicates the horizontal direction, residual of the sub-transform block of the specific layer horizontally divided is forcibly set to 0 and the sub-transform block thereof is not inversely transformed, and corresponding transform kernels including a discrete sine transform 7 (DST 7) transform kernel, a discrete cosine transform 8 (DCT 8) transform kernel, and a discrete cosine transform 2 (DCT 2) are applied to the sub-transform blocks of the other layers except the specific layer; or alternatively

Wherein when the division direction indicates the vertical direction, residuals of sub-transform blocks of a specific portion of the vertical division are forcibly set to 0 and sub-transform blocks thereof are not inversely transformed, and corresponding transform kernels including a discrete sine transform 7 (DST 7) transform kernel, a discrete cosine transform 8 (DCT 8) transform kernel, and a discrete cosine transform 2 (DCT 2) are applied to sub-transform blocks of other portions except the specific portion.

16. The method of claim 15, wherein the mode information of the current decoding block dividing the sub-transform block includes a division direction for indicating division of the current encoding block in a horizontal direction or a vertical direction and block number information indicating division of the current decoding block into the sub-transform blocks.

17. The method of claim 16, wherein parsing the mode information of the current decoding block divided sub-transform block and the coefficients of the sub-transform block from the code stream comprises:

parsing the identity of whether the current decoding block is divided into sub-transform blocks,

Parsing an identification indicating the number of blocks of the current decoding block divided into sub-transform blocks,

Resolving the partition direction of the current decoding block, and

The coefficients of the sub-transform block are parsed.

18. The method of claim 17, wherein parsing the identification indicating the number of blocks of the current decoding block divided into sub-transform blocks comprises:

When the width and the height of the current decoding block are smaller than a preset threshold value, the identification and the dividing direction of the block number which indicates that the current decoding block is divided into sub-transformation blocks are not analyzed; otherwise, resolving the identification indicating the number of blocks of the current decoding block divided into sub-transform blocks, and resolving the dividing direction if the number of blocks of the sub-transform blocks is greater than or equal to 3.

19. The method of claim 17, wherein resolving a partitioning direction comprises:

if the ratio of the width to the height of the current decoding block is greater than a preset threshold value or the ratio of the width to the height of the current decoding block is less than the preset threshold value, the dividing direction is not analyzed; otherwise, the dividing direction is analyzed.

20. The method of claim 17, wherein resolving a partitioning direction comprises:

after resolving the identification indicating the number of blocks of the current decoding block divided into sub-transform blocks, resolving the division direction if it is determined that the number of blocks of the divided sub-transform blocks is 3 or more.

21. The method of claim 15, wherein parsing coefficients of the sub-transform block comprises:

the flag CBF indicating whether or not the coefficient is contained in each sub-transform block is not parsed, and for the case where n=3, the coefficient of the first sub-transform block is default to 1, the coefficient of the second sub-transform block is 0, and the coefficient of the third sub-transform block is 1.

22. The method of claim 15, wherein parsing coefficients of the sub-transform block comprises:

The flag CBF indicating whether or not the coefficient is contained in each sub-transform block is not parsed, and for the case where n=3, the coefficient of the first sub-transform block is default to 0, the coefficient of the second sub-transform block is 1, and the coefficient of the third sub-transform block is 0.

23. The method of claim 16, wherein when the division direction indicates a horizontal direction and n=3, for the uppermost sub-transform block, inverse transform is performed by applying DST7 transform kernel in the horizontal direction and inverse transform is performed by applying DCT8 transform kernel in the vertical direction; forcing its residual to 0 for the middle sub-transform block and not performing the inverse transform; for the sub transform block of the lowest layer, the DST7 transform kernel is applied in the horizontal direction for inverse transform, and the DST7 transform kernel is applied in the vertical direction for inverse transform.

24. The method of claim 16, wherein inversely transforming coefficients of different sub-transform blocks corresponding to the sub-transform blocks comprises: when the division direction indicates a horizontal direction and n=3, residuals of the uppermost sub-transform block and the lowermost sub-transform block are forcedly set to 0, and the uppermost sub-transform block and the lowermost sub-transform block are not inversely transformed, and discrete cosine transform 2 (DCT 2) is employed as a transform kernel for the intermediate sub-transform block in both the horizontal direction and the vertical direction.

25. The method of claim 16, wherein the mode information of the current coding block partition sub-transform block comprises: when the division direction indicates a horizontal direction and n=3, the current coding block is horizontally divided into an uppermost sub-transform block having a height of current coding block height/4, an intermediate sub-transform block having a height of current coding block height/2, and a lowermost sub-transform block having a height of current coding block height/4.

26. The method of claim 16, wherein inversely transforming coefficients of different sub-transform blocks corresponding to the sub-transform blocks comprises: when the division direction indicates a vertical direction and n=3, for the leftmost sub-transform block, inverse transform is performed by applying a DCT8 transform kernel in a horizontal direction and inverse transform is performed by applying a DST7 transform kernel in a vertical direction; forcing its residual to 0 for the middle sub-transform block and not performing the inverse transform; for the rightmost sub-transform block, the DST7 transform kernel is applied in the horizontal direction for inverse transform, and the DST7 transform kernel is applied in the vertical direction for inverse transform.

27. The method of claim 16, wherein inversely transforming coefficients of different sub-transform blocks corresponding to the sub-transform blocks comprises:

when the division direction indicates a vertical direction and n=3, residuals of the leftmost sub-transform block and the rightmost sub-transform block are both forcibly set to 0, and the leftmost sub-transform block and the rightmost sub-transform block are not inversely transformed, and the middle sub-transform block is inversely transformed in both the horizontal direction and the vertical direction using DCT2 as a transform kernel.

28. The method of claim 16, wherein the mode information of the current coding block partition sub-transform block comprises: when the division direction indicates a vertical direction and n=3, the current encoding block is vertically divided into a leftmost sub-transform block having a width of current encoding block width/4, an intermediate sub-transform block having a width of current encoding block width/2, and a rightmost sub-transform block having a width of current encoding block width/4.

29. An apparatus for performing a method of transforming in video image encoding, comprising:

a transceiver configured to transmit and receive data; and

A controller coupled with the transceiver and configured to perform the method of any of claims 1-14.

30. An apparatus for performing a method of transforming in video image decoding, comprising:

a transceiver configured to transmit and receive data; and

A controller coupled with the transceiver and configured to perform the method of any of claims 15-28.

31. An apparatus for performing a method of transforming in video image encoding, the apparatus comprising:

A prediction unit configured to receive image data and to predict a predicted value of a current encoded block in the image data and to obtain a residual of the current encoded block from an original value of the image data;

A transform unit coupled to the prediction unit and configured to divide a current encoding block of image data into n sub-transform blocks according to mode information of the sub-transform blocks, and transform residuals of the generated different sub-transform blocks corresponding to the sub-transform blocks, and obtain coefficients of the sub-transform blocks by quantizing a transform result, and coefficient-encode the coefficients of the sub-transform blocks, wherein n is an integer greater than or equal to 3; and

A writing unit coupled to the transform unit and configured to write mode information of the current coding block dividing sub-transform blocks and a result of the coefficient coding into a code stream,

Wherein, when the division direction indicates the horizontal direction, residual errors of sub-transform blocks of a specific layer of the horizontal division are forcibly set to 0 and the sub-transform blocks thereof are not transformed, and corresponding transform kernels including a discrete sine transform 7 (DST 7) transform kernel and a discrete cosine transform 8 (DCT 8) transform kernel are applied to the sub-transform blocks of other layers except the specific layer; or alternatively

Wherein when the division direction indicates the vertical direction, residuals of sub-transform blocks of a specific portion of the vertical division are forcedly set to 0 and are not transformed, and corresponding transform kernels including a discrete sine transform 7 (DST 7) transform kernel and a discrete cosine transform 8 (DCT 8) transform kernel are applied to the sub-transform blocks of other portions except the specific portion.

32. The apparatus of claim 31, wherein the mode information of the current encoding block dividing the sub-transform block includes a division direction indicating division of the current encoding block in a horizontal direction or a vertical direction and block number information indicating division of the current encoding block into the sub-transform blocks.

33. The apparatus of claim 32, wherein the transform unit is configured to transform the generated residuals of different sub-transform blocks corresponding to the sub-transform block, comprising: when the division direction indicates a horizontal direction and n=3, for the uppermost sub-transform block, a discrete sine transform 7 (DST 7) transform kernel is applied in the horizontal direction and a discrete cosine transform 8 (DCT 8) transform kernel is applied in the vertical direction; for the middle sub-transform block, its residual is forced to 0 and no transform is performed; for the sub transform block of the lowest layer, DST7 transform kernels are applied in the horizontal direction, and DST7 transform kernels are applied in the vertical direction.

34. The apparatus of claim 32, wherein the mode information of the current coding block partition sub-transform block comprises: when the division direction indicates a horizontal direction and n=3, the current coding block is horizontally divided into an uppermost sub-transform block having a height of current coding block height/4, an intermediate sub-transform block having a height of current coding block height/2, and a lowermost sub-transform block having a height of current coding block height/4.

35. The apparatus of claim 32, wherein the transform unit is configured to transform the generated residuals of different sub-transform blocks corresponding to the sub-transform block, comprising: when the division direction indicates a vertical direction and n=3, for the leftmost sub transform block, a DCT8 transform kernel is applied in a horizontal direction and a DST7 transform kernel is applied in a vertical direction; forcing its residual to 0 for the middle sub-transform block and not transforming; for the rightmost sub-transform block, the DST7 transform kernel is applied in the horizontal direction and the DST7 transform kernel is applied in the vertical direction.

36. The apparatus of claim 32, wherein the mode information of the current coding block partition sub-transform block comprises: when the division direction indicates a vertical direction and n=3, the current encoding block is vertically divided into a leftmost sub-transform block having a width of current encoding block width/4, an intermediate sub-transform block having a width of current encoding block width/2, and a rightmost sub-transform block having a width of current encoding block width/4.

37. The apparatus of claim 31, wherein the writing unit configured to write mode information of the current encoded block divided sub-transform block and a result of the coefficient encoding to a code stream comprises:

Writing the identification of whether the current coding block needs to be divided into sub-transformation blocks into a code stream;

writing an identification indicating the number of blocks of the current encoded block divided into sub-transform blocks into a code stream;

Writing the dividing direction into the code stream; and

And writing the result of coefficient coding of the sub-transformation block into the code stream.

38. The apparatus of claim 37, wherein the writing unit is configured to write an identification indicating a number of blocks of the current encoded block divided into sub-transform blocks into a code stream, comprising:

When the width and the height of the current coding block are smaller than the preset threshold value, the code stream is not written with the identification indicating the number of blocks of the current coding block divided into sub-transformation blocks and the dividing direction, otherwise, the code stream is written with the identification indicating the number of blocks of the current coding block divided into sub-transformation blocks, and if the number of blocks of the divided sub-transformation blocks is larger than or equal to 3, the code stream is written with the dividing direction.

39. The apparatus of claim 37, wherein the writing unit is configured to write a partitioning direction to a code stream, comprising:

if the ratio of the width to the height of the current coding block is greater than a predetermined threshold value or the ratio of the width to the height of the current coding block is less than the predetermined threshold value, the dividing direction is not transmitted, otherwise the dividing direction is written into the code stream.

40. The apparatus of claim 37, wherein the transform unit is configured to coefficient encode coefficients of the sub-transform block, comprising:

The flag CBF of whether or not the coefficient is contained in each sub-transform block is not written, and for the case where n=3, the first sub-transform block coefficient is defaulted to 1, the second sub-transform block coefficient is 0, and the third sub-transform block coefficient is 1.

41. A device for performing a method of transforming in video image decoding, the device comprising: a parsing unit configured to receive the code stream and to parse mode information of the current decoding block dividing the sub-transform block and coefficients of the sub-transform block;

A transform unit coupled to the parsing unit and configured to divide a current decoding block into n sub-transform blocks according to mode information of the sub-transform blocks divided by the current encoding block, and perform inverse transform corresponding to the sub-transform blocks on coefficients of different sub-transform blocks to obtain residual errors of the sub-transform blocks, wherein n is an integer greater than or equal to 3; and

A reconstruction unit coupled to the transform unit and configured to reconstruct an image through a residual of a sub-transform block of the current decoding block and a prediction value of the current decoding block,

Wherein, when the division direction indicates the horizontal direction, residual errors of sub-transform blocks of a specific layer of the horizontal division are forcibly set to 0 and the sub-transform blocks thereof are not inversely transformed, and corresponding transform kernels including a discrete sine transform 7 (DST 7) transform kernel and a discrete cosine transform 8 (DCT 8) transform kernel are applied to the sub-transform blocks of other layers except the specific layer; or alternatively

Wherein when the division direction indicates the vertical direction, residuals of sub-transform blocks of a specific portion of the vertical division are forcibly set to 0 and sub-transform blocks thereof are not inversely transformed, and corresponding transform kernels including a discrete sine transform 7 (DST 7) transform kernel and a discrete cosine transform 8 (DCT 8) transform kernel are applied to sub-transform blocks of other portions except the specific portion.

42. The apparatus of claim 41, wherein the mode information of the current decoding block dividing the sub-transform block includes a division direction indicating division of the current encoding block in a horizontal direction or a vertical direction and block number information indicating division of the current decoding block into the sub-transform blocks.

43. The apparatus of claim 42, wherein the parsing unit is configured to parse mode information of a current decoding block partition sub-transform block and coefficients of the sub-transform block from a code stream, comprising:

parsing the identity of whether the current decoding block is divided into sub-transform blocks,

Parsing an identification indicating the number of blocks of the current decoding block divided into sub-transform blocks,

Resolving the partition direction of the current decoding block, and

The coefficients of the sub-transform block are parsed.

44. The apparatus of claim 43, wherein the parsing unit is configured to parse an identification indicating a number of blocks of the current decoding block divided into sub-transform blocks, comprising: when the width and the height of the current decoding block are smaller than a preset threshold value, the identification and the dividing direction of the block number which indicates that the current decoding block is divided into sub-transformation blocks are not analyzed; otherwise, resolving the identification indicating the number of blocks of the current decoding block divided into sub-transform blocks, and resolving the dividing direction if the number of blocks of the sub-transform blocks is greater than or equal to 3.

45. The apparatus of claim 43, wherein the parsing unit is configured to parse a partitioning direction, comprising: if the ratio of the width to the height of the current decoding block is greater than a preset threshold value or the ratio of the width to the height of the current decoding block is less than the preset threshold value, the dividing direction is not analyzed; otherwise, the dividing direction is analyzed.

46. The apparatus of claim 41, wherein the parsing unit is configured to parse coefficients of a sub-transform block, comprising:

the flag CBF indicating whether or not the coefficient is contained in each sub-transform block is not parsed, and for the case where n=3, the coefficient of the first sub-transform block is default to 1, the coefficient of the second sub-transform block is 0, and the coefficient of the third sub-transform block is 1.

47. An apparatus as defined in claim 42, wherein the transform unit is configured to inverse transform coefficients of different sub-transform blocks corresponding to the sub-transform blocks, comprising: when the division direction indicates a horizontal direction and n=3, for the uppermost sub-transform block, inverse transform is performed by applying a discrete sine transform 7 (DST 7) transform kernel in the horizontal direction and inverse transform is performed by applying a discrete cosine transform 8 (DCT 8) transform kernel in the vertical direction; forcing its residual to 0 for the middle sub-transform block and not performing the inverse transform; for the sub transform block of the lowest layer, the DST7 transform kernel is applied in the horizontal direction for inverse transform, and the DST7 transform kernel is applied in the vertical direction for inverse transform.

48. An apparatus as defined in claim 42, wherein the mode information of the current coding block partition sub-transform block comprises: when the division direction indicates a horizontal direction and n=3, the current coding block is horizontally divided into an uppermost sub-transform block having a height of current coding block height/4, an intermediate sub-transform block having a height of current coding block height/2, and a lowermost sub-transform block having a height of current coding block height/4.

49. An apparatus as defined in claim 42, wherein the transform unit is configured to inverse transform coefficients of different sub-transform blocks corresponding to the sub-transform blocks, comprising: when the division direction indicates a vertical direction and n=3, for the leftmost sub-transform block, inverse transform is performed by applying a DCT8 transform kernel in a horizontal direction and inverse transform is performed by applying a DST7 transform kernel in a vertical direction; forcing its residual to 0 for the middle sub-transform block and not performing the inverse transform; for the rightmost sub-transform block, the DST7 transform kernel is applied in the horizontal direction for inverse transform, and the DST7 transform kernel is applied in the vertical direction for inverse transform.

50. An apparatus as defined in claim 42, wherein the mode information of the current coding block partition sub-transform block comprises: when the division direction indicates a vertical direction and n=3, the current encoding block is vertically divided into a leftmost sub-transform block having a width of current encoding block width/4, an intermediate sub-transform block having a width of current encoding block width/2, and a rightmost sub-transform block having a width of current encoding block width/4.