JP7166743B2 - Encoding device, decoding device and program - Google Patents

Description

The present invention relates to an encoding device, a decoding device and a program.

H.265/HEVC (High Efficiency Video Coding)
In the moving image (video) coding method represented by , prediction is performed while switching between two types of prediction, inter prediction using temporal correlation between frames and intra prediction using spatial correlation within a frame. After generating a difference signal, it is configured to output a stream obtained by performing orthogonal transform processing, loop filter processing, and entropy coding processing.

In intra prediction in HEVC, a total of 35 predictions such as Planer prediction, DC prediction, and directional prediction
Various modes are prepared, and intra prediction is performed using adjacent decoded reference pixels according to the mode determined by the encoder.

Here, in intra prediction, a specified value (10 It is configured to create reference pixels to be used when generating a predicted image by filling in "512" in the case of a bit moving image).

In addition, in conventional HEVC, encoding processing is performed in raster scan order from the upper left, so there are cases where reference pixels have not been decoded. In such a case, a prediction image is generated using a value obtained by extrapolating the nearest decoded reference pixel to the 0th order.

In particular, in intra prediction in conventional HEVC, reference pixels located at the lower left and upper right of the CU have already been decoded by the encoding process in the raster scan order shown in FIG. In many cases, it is not (see FIG. 6B), and in such a case, if directional prediction is performed from the direction in which there is a reference pixel that has not been decoded, the prediction accuracy will decrease, and the coding efficiency will decrease. There was a problem.

In order to solve such a problem, in intra prediction, raster scan order (for example, Z type) is used as the encoding processing order for a plurality of transform blocks existing in a CU (hereinafter referred to as "TU: Transform Unit"). In addition, there is known a technique for improving prediction accuracy by giving flexibility to the order of encoding such as U type and X type (see Non-Patent Document 1).
.

In the examples of FIGS. 6A and 6B, direction prediction is performed in the direction from the lower left to the upper right (the direction opposite to the direction indicated by the dashed arrows in FIGS. 6A and 6B). The pixel on the dashed arrow is predicted using the lower left reference pixel. In the drawings of this specification, the arrow indicating the direction of the intra prediction mode (prediction direction) points from the target pixel for intra prediction to the reference pixel, as described in the HEVC standard (the same applies hereinafter).

Intra-prediction and inter-prediction are used for patterns such as when texture exists in a very localized area in a flat image, or when a small object moves in a different direction in a global motion that moves the entire screen. The prediction accuracy is high in many regions of the predicted image, and the energy of the residual signal is concentrated in some regions where the prediction was not successful.

In this way, when the energy of the residual signal is already concentrated at the stage before the orthogonal transform process is applied, the application of the orthogonal transform process may increase the entropy.

Therefore, in the conventional HEVC, "Tran
sformSkip mode” is provided.

In conventional HEVC, as shown in FIG. 7, intra prediction is prediction using spatially adjacent upper or left decoded reference pixels, and the accuracy of the predicted image at a position close to the decoded reference pixels is high. Predictive images at positions far from decoded reference pixels tend to be less accurate.

That is, in the residual signal generated by intra prediction, the energy tends to be low at positions close to the decoded reference pixels and high at positions far from the decoded reference pixels.

When orthogonal transform processing is applied to the residual signal, transform coefficients obtained by orthogonal transform processing tend to have large energy in the upper left region corresponding to low frequency components.

This tendency is also used in the entropy coding process, but Transform
In the mSkip mode, since the energy of the residual signal increases as the position is farther from the decoded reference pixels, in the case of intra prediction using the left or upper reference pixels, the energy increases in the lower right area of the residual signal. tend to become

Therefore, in the TransformSkip mode, if entropy coding processing similar to the case where the orthogonal transform processing is applied to the residual signal is applied, there is a problem that the coding efficiency is lowered.

Therefore, in order to solve this problem, in HEVC, when the TransformSkip mode is applied, the energy distribution of the residual signal is inverted vertically and horizontally so that the orthogonal transform process is applied. It approximates the energy distribution of the transform coefficients generated when the entropy encoding process is performed, and improves the efficiency of the subsequent entropy coding process.

Mochizuki et al., "Applied intra-prediction method based on mean value coordinates", Information Processing Society of Japan research report, vol, 2012-AVM-77, No.12

As described above, in the technique described in Non-Patent Document 1, there are cases where reference pixels on the lower side or the right side are used in intra prediction. In such a case, the residual signal tends to have low signal intensity on the lower side and right side closer to the reference pixel position, and higher signal intensity on the upper side and left side farther from the reference pixel position.

However, in the TransformSkip mode in conventional HEVC, the residual signal is vertically and horizontally inverted regardless of the position of the reference pixel. Then the energy of the residual signal is
There is a problem that the coding efficiency is lowered because the coding efficiency is not concentrated in the upper left area.

Therefore, the present invention has been made to solve the above-described problems, and is a case where at least one of the lower and right reference pixels is used in intra prediction and the residual signal is subjected to orthogonal transform processing (or It is an object of the present invention to provide an encoding device, a decoding device, and a program that can suppress a decrease in encoding efficiency even when inverse orthogonal transform processing is not applied.

A first feature of the present invention is an encoding device that is configured to divide and encode an original image in units of frames that constitute a moving image into encoding target blocks, and uses an intra prediction mode to encode An intra prediction unit configured to generate a predicted image, and a residual signal generation unit configured to generate a residual signal from a difference between the predicted image and the original image, When the orthogonal transform processing is not applied, the residual signal generation unit converts the residual signal in at least the horizontal direction and the vertical direction based on the positions of the reference pixels used when the intra prediction unit generates the predicted image. The gist is that it is configured to be reversed to one side.

A second feature of the present invention is a decoding device configured to divide an original image in units of frames constituting a moving image into encoding target blocks and decode the predicted image using an intra prediction mode. and an intra prediction unit configured to generate an entropy decoding process and an inverse an inverse transform unit configured to generate a residual signal by horizontally and/or vertically inverting a signal resulting from the quantization process.

A gist of a third aspect of the present invention is a program for causing a computer to function as the encoding device according to the first aspect.

A gist of a fourth aspect of the present invention is a program for causing a computer to function as the decoding device according to the second aspect.

According to the present invention, even when at least one of the lower and right reference pixels are used in intra prediction and orthogonal transform processing is not applied to the residual signal, a decrease in coding efficiency can be suppressed. It is possible to provide an encoding device, a decoding device, and a program that can

FIG. 1 is a functional block diagram of an encoding device 1 according to the first embodiment. FIG. 2 is a diagram illustrating an example of intra prediction when TU partitioning is performed in the first embodiment. FIG. 3 is a flow chart showing the operation of the encoding device 1 according to the first embodiment. FIG. 4 is a functional block diagram of the decoding device 3 according to the first embodiment. FIG. 5 is a flow chart showing the operation of the decoding device 3 according to the first embodiment. FIG. 6 is a diagram for explaining conventional HEVC. FIG. 7 is a diagram for explaining conventional HEVC.

(First embodiment)
An encoding device 1 and a decoding device 3 according to the first embodiment of the present invention will be described below with reference to FIGS. 1 to 5. FIG. Here, the encoding device 1 and the decoding device 3 according to the present embodiment are HE
It is configured to support intra-prediction in moving image coding methods such as VC. Note that the encoding device 1 and the decoding device 3 according to the present embodiment are configured to be compatible with any video encoding method as long as it is a video encoding method that performs intra prediction.

The encoding device 1 according to the present embodiment is configured to divide an original image in units of frames constituting a moving image into CUs and encode the CUs. Further, the encoding device 1 according to the present embodiment has a CU
can be divided into multiple TUs. Hereinafter, in this embodiment, a case in which a CU is divided into a plurality of TUs will be described as an example, but the present invention can also be applied to a case in which a CU is not divided into a plurality of TUs.

Note that, in this embodiment, a CU to be encoded that does not have an adjacent decoded reference pixel, such as the CU located at the upper leftmost position in a frame, is assigned a specified value (for a 10-bit video, "
512”) is configured to generate reference pixels used when generating a predicted image, all pixels adjacent to the left side of the CU to be coded can be used as reference pixels. do.

As shown in FIG. 1, the encoding device 1 according to this embodiment includes an intra prediction mode determination unit 11
, TU division determination unit 12 , encoding order control unit 13 , sequential local decoded image generation unit 14 , memory 15 , and entropy encoding unit 16 .

The intra-prediction mode determination unit 11 is configured to determine the optimum intra-prediction mode to apply to the CU.

The TU partition decision unit 12 is configured to decide whether to partition a CU into a plurality of TUs. In this embodiment, as a method of dividing a CU into a plurality of TUs, a case of 4 divisions is described as an example. , is not limited to such cases.

The coding order control unit 13 is configured to determine the coding order of the TUs within the CU based on the intra prediction mode (for example, the direction of the intra prediction mode).

Specifically, when the TU division determination unit 12 determines to divide a CU into a plurality of TUs, the coding order control unit 13 performs the following operations as shown in FIGS. , when the direction of the intra prediction mode determined by the intra prediction mode determination unit 11 is the direction from the lower left to the upper right (that is, when the direction prediction is performed from the lower left to the upper right), the CU
As the coding order of TUs in the conventional raster scan order (Z type as shown in FIG. 6(a))
TU#A3 (lower left TU in CU#A) → TU#A4 (lower right TU in CU#A
)→TU#A1 (upper left TU in CU#A)→TU#A2 (upper right TU in CU#A), or TU#A3 (lower left TU in CU#A)→ TU#A1 (CU#A
(upper left TU in CU#A) → TU#A4 (lower right TU in CU#A) → TU#A2 (upper right TU in CU#A). may be configured to

Further, the coding order control unit 13 determines that the TU division determination unit 12 determines to divide the CU into a plurality of TUs, and the direction of the intra prediction mode determined by the intra prediction mode determination unit 11 is TU# A2 (upper right TU in CU#A) → TU#A4 (CU#A
lower right TU in CU#A) → TU#A1 (upper left TU in CU#A) → TU#A3 (lower left TU in CU#A), or TU#A2 (in CU#A upper right TU) → TU#
A1 (upper left TU in CU#A)→TU#A4 (lower right TU in CU#A)→TU#A3
It may be configured to adopt a predetermined encoding order among the encoding orders of (lower left TU in CU#A).

The sequential locally decoded image generation unit 14 is configured to generate a locally decoded image (a decoded image for each TU) based on the coding order determined by the coding order control unit 13 and the method of dividing the CU into TUs. ing.

Specifically, when the TU division determination unit 12 determines to divide the CU into a plurality of TUs, the sequential locally decoded image generation unit 14 follows the coding order determined by the coding order control unit 13. , is configured to sequentially generate a locally decoded image.

As shown in FIG. 1, the sequential local decoded image generation unit 14 includes an intra prediction unit 14a, a residual signal generation unit 14b, an orthogonal transform/quantization unit 14c, and an inverse quantization/inverse orthogonal transformation unit 14d. ,
and a locally decoded image generator 14e.

The intra prediction unit 14 a is configured to generate a predicted image using the intra prediction mode determined by the intra prediction mode determination unit 11 . That is, the intra prediction unit 14d is configured to determine the positions of reference pixels used when generating a predicted image.

Specifically, the intra prediction unit 14a divides the CU into a plurality of TUs by the TU partition determination unit 12.
When it is decided to divide into, and as shown in FIGS. 2A to 2D, the direction of the intra prediction mode (prediction direction) is the direction from the lower left to the upper right, TU#A3 (CU
# Lower left TU in A) → TU#A4 (Lower right TU in CU#A) → TU#A1 (CU#A
upper left TU in) → TU #A2 (upper right TU in CU #A), or T
U#A3 (bottom left TU in CU#A)→TU#A1 (top left TU in CU#A)→TU#
It is configured to generate a prediction image in a predetermined coding order among the coding order of A4 (lower right TU in CU#A)→TU#A2 (upper right TU in CU#A). may be

Here, as shown in FIGS. 2(c) and 2(d), the intra prediction unit 14a predicts TU#A1 (upper left TU in CU#A) whose adjacent lower reference pixel is decoded. and TU#A2
For (the upper right TU in CU#A), the predicted image may be generated using the decoded reference pixels adjacent to the left and lower sides.

In addition, in the encoding device 1 according to the present embodiment, the intra prediction unit 14a performs the direction of the intra prediction mode (prediction direction) is from top right to bottom left, then TU#A2 (CU#
A) → TU#A4 (lower right TU within CU#A) → TU#A1 (upper left TU within CU#A) → TU#A3 (lower left TU within CU#A) or TU
#A2 (upper right TU in CU#A)→TU#A1 (upper left TU in CU#A)→TU#A
4 (lower right TU in CU#A) → TU#A3 (lower left TU in CU#A). may be

Here, the intra prediction unit 14a predicts TU#A in which the adjacent right reference pixel is decoded.
For 1 (upper left TU in CU#A) and TU#A3 (lower left TU in CU#A),
The predicted image may be generated using the upper and right adjacent decoded reference pixels.

Alternatively, the intra prediction unit 14a performs TU#A3 (the lower left TU in CU#A)→TU#A
4 (bottom right TU in CU#A)→TU#A1 (top left TU in CU#A)→TU#A2 (
upper right TU in CU#A), or TU#A3 (lower left T in CU#A)
U) → TU#A1 (top left TU in CU#A) → TU#A4 (bottom right TU in CU#A)
→ When a coding order of TU#A2 (upper right TU in CU#A) is used, a TU whose reference pixel adjacent to the upper side has been decoded (the uppermost TU among the divided TU groups TU
, TU#A1 and TU#A2 in the example of FIG. 2) are not in the common intra-prediction direction in CU#A, but the decoded reference pixels adjacent to the left side, the upper side, or the lower side of the TU are used. It may be configured to perform a predefined prediction, such as linear interpolation.

The residual signal generation unit 14b is configured to generate a residual signal from the difference between the predicted image generated by the intra prediction unit 14a and the original image.

Here, when the "TransformSkip mode" is applied (when the orthogonal transform is not applied), the residual signal generation unit 14b uses reference pixels to generate the predicted image determined by the intra prediction unit 14a. may be configured to invert the generated residual signal in at least one of the horizontal and vertical directions based on the position of .

For example, when the "TransformSkip mode" is applied and the intra prediction unit 14a generates a predicted image using the reference pixels located on the left side and the lower side, the residual signal generation unit 14b generates The residual signal may be horizontally inverted to concentrate the energy of the residual signal in the upper left region.

Further, when the "TransformSkip mode" is applied and the intra prediction unit 14a generates a predicted image using the reference pixels located on the right side and the upper side,
The residual signal generator 14b may be configured to vertically invert the generated residual signal and concentrate the energy of the residual signal in the upper left region.

Further, when the "TransformSkip mode" is applied and the intra prediction unit 14a generates a predicted image using the reference pixels located on the left side and the upper side,
Similar to the case of applying the “TransformSkip mode” in conventional HEVC, the residual signal generation unit 14b vertically and horizontally inverts the generated residual signal, and concentrates the energy of the residual signal in the upper left region. may be configured to

Further, when the "TransformSkip mode" is applied, and when the intra prediction unit 14a generates a predicted image using the reference pixels located on the left side, the upper side, and the lower side, the residual signal generation unit 14b , may be configured to horizontally invert the generated residual signal, concentrating the energy of the residual signal in the left region.

Furthermore, when the "TransformSkip mode" is applied, and when the intra prediction unit 14a generates a predicted image using the reference pixels located on the left side, the upper side, and the right side, the residual signal generation unit 14b The generated residual signal may be vertically inverted to concentrate the energy of the residual signal in the upper region.

Note that the residual signal inversion process may be realized by another function such as the orthogonal transformation/quantization unit 14c or the entropy coding unit 16 instead of the residual signal generation unit 14b.
For example, instead of inverting the residual signal, the entropy coding unit 16 may change the scan order of the coefficients to obtain the same effect as inverting the residual signal.

On the other hand, when the "Transform Skip mode" is applied, and when the intra prediction unit 14a does not generate a predicted image using reference pixels located on either the left side or the upper side, the generated residue It may be configured not to invert the difference signal.

The orthogonal transformation/quantization unit 14c is configured to perform orthogonal transformation processing and quantization processing on the residual signal generated by the residual signal generation unit 14b, and generate quantized transform coefficients. .

Here, when the "TransformSkip mode" is applied, the orthogonal transformation/quantization unit 14c quantizes the residual signal generated by the residual signal generation unit 14b without performing orthogonal transformation processing. configured for processing only.

Alternatively, when the "TransformSkip mode" is applied, neither the orthogonal transform processing nor the quantization processing is performed, and the residual signal generated by the residual signal generation unit 14b is output as it is. may

The inverse quantization unit/inverse orthogonal transform unit 14d performs inverse quantization processing and inverse orthogonal transform processing again on the quantized transform coefficients generated by the orthogonal transform/quantization unit 14c to generate a residual signal. configured to generate

Here, the inverse quantization unit/inverse orthogonal transform unit 14d uses "TransformSkip mode"
is applied, the quantized residual signal generated by the orthogonal transform/quantization unit 14c is not subjected to the inverse orthogonal transform process, but only the inverse quantization process is performed to obtain the residual signal is configured to generate

Alternatively, when the "TransformSkip mode" is applied, neither the orthogonal transform processing nor the quantization processing is performed, and the residual signal generated by the orthogonal transform/quantization unit 14c is configured to be output as it is. may be

The local decoded image generation unit 14e generates a local decoded image by adding the predicted image generated by the intra prediction unit 14a to the residual signal generated by the inverse quantization unit/inverse orthogonal transformation unit 14d. It is configured.

The memory 15 is configured to retain locally decoded images generated by the sequential local decoded image generator 14 so as to be usable as reference images.

The entropy coding unit 16 is configured to perform entropy coding processing on the flag information including the intra prediction mode determined by the intra prediction mode determination unit 11 and the quantized transform coefficients, and to output a stream. there is

FIG. 3 shows a flowchart for explaining an example of the operation of the encoding device 1 according to this embodiment when the "TransformSkip mode" is applied.

As shown in FIG. 3, in step S101, the encoding device 1 determines the optimum intra prediction mode to apply to the CU.

In step S102, the encoding device 1 determines whether to divide the CU into multiple TUs. If it is determined in step S102 that the CU is to be split into multiple TUs, the operation proceeds to step S103. On the other hand, if it is determined in step S102 that the CU is not split into multiple TUs, the operation proceeds to step S108.

When it is determined in step S103 that the direction of the intra prediction mode is the direction from the lower left to the upper right or the direction from the upper right to the lower left, this operation proceeds to step S10.
Proceed to 5. On the other hand, when it is determined in step S103 that the direction of the intra prediction mode is other than the direction from the lower left to the upper right and the direction from the upper right to the lower left, the operation proceeds to step S104.

In step S104, the encoding device 1 uses conventional HEVC as the encoding order described above.
adopt the raster scan order (Z type as shown in FIG. 8(a)) used in .

In step S108, the encoding device 1 encodes the TU to be encoded.
A predefined prediction is performed using the neighboring decoded reference pixels to the left and top of .

When it is determined that the direction of the intra prediction mode is from the lower left to the upper right (step S105), in step S106, the encoding device 1 sets TU#A3 (CU#A lower left TU in CU#A) → TU#A4 (lower right TU in CU#A) →
TU#A1 (top left TU within CU#A)→TU#A2 (top right TU within CU#A), or TU#A3 (bottom left TU within CU#A)→TU# A1 (upper left TU in CU#A) → TU#A4 (lower right TU in CU#A) → TU#A2 (upper right TU in CU#A) Adopt encoding order.

On the other hand, when it is determined that the direction of the intra prediction mode is not the direction from the lower left to the upper right (step S105), in step S111, the encoding device 1 sets TU#A2 (CU# Upper right TU in A) → TU#A4 (T at lower right in CU#A
U)→TU#A1 (upper left TU in CU#A)→TU#A3 (lower left TU in CU#A)
or TU#A2 (upper right TU in CU#A) → TU#A1 (CU#
TU#A4 (lower right TU in CU#A)→TU#A3 (lower left TU in CU#A). adopt.

In step S107, the encoding device 1 determines whether or not the reference pixel adjacent to the upper side of the TU to be encoded has been decoded. In step S107, if the decoding has been completed, the operation proceeds to step S109, and if the decoding has not been completed, the operation proceeds to step S110.

In step S109, the encoding device 1 encodes the TU to be encoded.
A predefined prediction is performed using the neighboring decoded reference pixels to the left and above and below of .

In step S110, the encoding device 1 encodes the TU to be encoded.
A predefined prediction is performed using the neighboring decoded reference pixels to the left and to the bottom of .

In step S112, the encoding device 1 determines whether or not the reference pixel adjacent to the left side of the TU to be encoded has been decoded. In step S112, if the decoding has been completed, the operation proceeds to step S113, and if the decoding has not been completed, the operation proceeds to step S114.

In step S113, the encoding device 1 encodes the TU to be encoded.
A predefined prediction is performed using the neighboring decoded reference pixels to the left, top and right of .

In step S114, the encoding device 1 encodes the TU to be encoded.
A predefined prediction is performed using the decoded reference pixels that are adjacent to the right and top of .

In step S115, the encoding device 1 vertically inverts the residual signal generated using the predicted image and the original image, and then performs subsequent processing.

In step S116, the encoding device 1 horizontally inverts the residual signal generated using the predicted image and the original image, and then performs subsequent processing.

In step S117, the encoding device 1 vertically and horizontally inverts the residual signal generated using the predicted image and the original image, and then performs subsequent processing.

According to the encoding device 1 according to the present embodiment, when at least one of the reference pixels on the lower side and the right side is used in intra prediction and the TransformSkip mode is applied (orthogonal transform processing is performed on the residual signal). is not applied), it is possible to suppress the deterioration of the coding efficiency.

Further, the decoding device 3 according to the present embodiment converts the original image of each frame constituting the moving image into a CU
It is configured to be divided into and decoded. Also, the decoding device 3 according to this embodiment is configured to be able to divide a CU into a plurality of TUs, like the encoding device 1 according to this embodiment.

As shown in FIG. 4, the decoding device 3 according to this embodiment includes an entropy decoding unit 31, a decoding order control unit 32, a sequential local decoded image generation unit 33, and a memory .

The entropy decoding unit 31 is configured to decode transform coefficients, flag information, and the like from the stream output from the encoding device 1 by performing entropy decoding processing on the stream output from the encoding device 1. ing. Here, the transform coefficients are the encoding device 1
is a quantized transform coefficient obtained as a signal encoded by dividing an original image in units of frames into CUs.

The flag information also includes accompanying information such as information indicating whether or not the prediction mode or the "TransformSkip mode" is selected.

The decoding order control unit 32 is configured to determine the decoding order of TUs within the CU based on the intra prediction mode.

Specifically, the decoding order control unit 32 uses a flag indicating whether or not the TU division output by the entropy decoding unit 31 has been performed (whether or not the CU is divided into a plurality of TUs) and the intra prediction mode. The direction is configured to determine the decoding order of the TUs within the CU.

For example, similar to the encoding order control unit 13, the decoding order control unit 32, when the CU is divided into a plurality of TUs and the direction of the intra prediction mode is from the lower left to the upper right, TU#A3 (lower left TU within CU#A) → TU#A4 (lower right TU within CU#A)
→ TU#A1 (upper left TU in CU#A) → TU#A2 (upper right TU in CU#A), or TU#A3 (lower left TU in CU#A) → TU# Predefined decoding in the decoding order of A1 (upper left TU in CU#A) → TU#A4 (lower right TU in CU#A) → TU#A2 (upper right TU in CU#A) It may be configured to perform decoding processing in order.

Further, similarly to the encoding order control unit 13, the decoding order control unit 32, when the CU is divided into a plurality of TUs and the direction of the intra prediction mode is from the upper right to the lower left, TU#A2 (top right TU within CU#A)→TU#A4 (bottom right TU within CU#A)→
The decoding order of TU#A1 (top left TU within CU#A)→TU#A3 (bottom left TU within CU#A), or TU#A2 (top right TU within CU#A)→TU#A1 Predetermined decoding in the decoding order of (upper left TU in CU#A) → TU#A4 (lower right TU in CU#A) → TU#A3 (lower left TU in CU#A) It may be configured to perform decoding processing in order.

The sequential local decoded image generation unit 33 uses the decoding order and the CU determined by the decoding order control unit 32.
is configured to generate a locally decoded image (a decoded image for each TU) based on the method of dividing into TUs.

Specifically, when the CU is divided into a plurality of TUs, the sequential locally decoded image generation unit 33 performs the quantization output by the entropy decoding unit 31 according to the decoding order determined by the decoding order control unit 32. A locally decoded image is generated by sequentially performing intra prediction, inverse quantization processing, and inverse orthogonal transform processing on the transformed transform coefficients.

As shown in FIG. 4, the sequential local decoded image generation unit 33 includes an intra prediction unit 33a, an inverse quantization/inverse transform unit 33b, and a decoded image generation unit 33c.

The intra prediction unit 33 a may be configured to generate a prediction image using the intra prediction mode output by the entropy decoding unit 31 according to the decoding order determined by the decoding order control unit 32 .

Specifically, when the CU is divided into a plurality of TUs and the direction of the intra prediction mode is from the lower left to the upper right, the intra prediction unit 33a determines TU #A3 (C
lower left TU in U#A) → TU#A4 (lower right TU in CU#A) → TU#A1 (CU#
A (upper left TU in A) → TU #A2 (upper right TU in CU #A), or T
U#A3 (bottom left TU in CU#A)→TU#A1 (top left TU in CU#A)→TU#
It is configured to generate a prediction image in a predetermined decoding order out of the decoding order of A4 (lower right TU in CU#A)→TU#A2 (upper right TU in CU#A). good too.

Here, as shown in FIGS. 2(c) and 2(d), the intra prediction unit 33a predicts TU#A1 (upper left TU in CU#A) whose adjacent lower reference pixel is decoded. and TU#A2
For (the upper right TU in CU#A), the predicted image may be generated using the decoded reference pixels adjacent to the left and lower sides.

Further, in the decoding device 3 according to the present embodiment, the intra prediction unit 33a can
and the direction of intra prediction mode (prediction direction) is from upper right to lower left, TU#A2 (upper right TU in CU#A) → TU#A4 (CU#
A) → TU#A1 (upper left TU in CU#A) → TU#A3 (lower left TU in CU#A), or TU#A2 (inside CU#A) upper right TU) → TU#
A1 (upper left TU in CU#A)→TU#A4 (lower right TU in CU#A)→TU#A3
The prediction image may be generated in a predetermined decoding order out of the decoding order (lower left TU in CU#A).

Here, the intra prediction unit 33a predicts TU#A in which the adjacent right reference pixel is decoded.
For 1 (upper left TU in CU#A) and TU#A3 (lower left TU in CU#A),
The prediction image may be generated using decoded reference pixels adjacent to the upper side and the right side.

Alternatively, the intra prediction unit 33a performs TU#A3 (the lower left TU in CU#A)→TU#A
4 (bottom right TU in CU#A)→TU#A1 (top left TU in CU#A)→TU#A2 (
upper right TU in CU#A), or TU#A3 (lower left TU in CU#A)
)→TU#A1 (upper left TU in CU#A)→TU#A4 (lower right TU in CU#A)→
When the decoding order of TU#A2 (upper right TU in CU#A) is used, the TU whose upper adjacent reference pixel has already been decoded (the TU located at the top of the divided TU group, In the example of FIG. 2, for TU#A1 and TU#A2), linear prediction using decoded reference pixels adjacent to the left side, upper side, or lower side of the TU is not performed in the common intra prediction direction in CU#A. It may be configured to perform predefined predictions such as interpolation.

The inverse quantization/inverse transform unit 33b performs inverse quantization processing and inverse transform processing (for example, inverse orthogonal transform processing) on the quantized transform coefficients output from the entropy decoding unit 31, thereby generating a residual configured to generate a signal;

Here, when the "TransformSkip mode" is applied (when the inverse orthogonal transform process is not applied), the inverse quantization/inverse transform unit 33b performs the inverse transform process on the signal obtained by the entropy decoding process. is configured to perform only inverse quantization processing without performing

Alternatively, when the “TransformSkip mode” is applied, neither the inverse orthogonal transform processing nor the inverse quantization processing may be performed, and the signal obtained by the entropy decoding processing may be output as it is.

Further, when the "TransformSkip mode" is applied, the inverse quantization/inverse transform unit 33b performs the entropy decoding process and the inverse based on the position of the reference pixel used when the intra prediction unit 33a generates the predicted image. The residual signal may be generated by horizontally and/or vertically inverting the signal obtained by the quantization process.

For example, when the "TransformSkip mode" is applied and the intra prediction unit 33a generates a predicted image using the reference pixels located on the left side and the lower side, the inverse quantization/inverse transform unit 33b , the residual signal may be generated by horizontally inverting the signal obtained by the entropy decoding process and the inverse quantization process.

Further, when the "TransformSkip mode" is applied and the intra prediction unit 33a generates a predicted image using the reference pixels located on the right side and the upper side,
The inverse quantization/inverse transform unit 33b may be configured to generate a residual signal by vertically inverting the signal obtained by the entropy decoding process and the inverse quantization process.

Further, when the "TransformSkip mode" is applied and the intra prediction unit 33a generates a predicted image using the reference pixels located on the left side and the upper side,
As in the case of applying the "TransformSkip mode" in conventional HEVC, the inverse quantization/inverse transform unit 33b vertically and horizontally inverts the signal obtained by the entropy decoding process and the inverse quantization process. It may be configured to generate a residual signal.

Further, when the "TransformSkip mode" is applied and the intra prediction unit 33a generates a predicted image using the reference pixels located on the left side, the upper side, and the lower side, the inverse quantization/inverse transform unit 33b may be configured to generate a residual signal by horizontally inverting the signal resulting from the entropy decoding and dequantization processes.

Furthermore, when the "TransformSkip mode" is applied and the intra prediction unit 33a generates a predicted image using the reference pixels located on the left, upper and right sides, the inverse quantization/inverse transform unit 33b may be configured to generate a residual signal by vertically inverting the signal resulting from the entropy decoding and inverse quantization processes.

On the other hand, when the “TransformSkip mode” is applied and when the intra prediction unit 33a does not generate a predicted image using reference pixels located on either the left side or the upper side, the inverse quantization/ The inverse transform unit 33b may be configured not to invert the signal obtained by the entropy decoding process and the inverse quantization process.

The decoded image generator 33c is configured to generate a local decoded image by adding the predicted image generated by the intra prediction unit 33a and the residual signal generated by the inverse quantization/inverse transform unit 33b. .

The memory 34 is configured to hold locally decoded images generated by the sequential locally decoded image generator 33 so as to be usable as reference images for intra prediction and inter prediction.

FIG. 5 shows a flowchart for explaining an example of the operation of determining the above-described decoding order by the decoding device 3 according to this embodiment when the "TransformSkip mode" is applied.

As shown in FIG. 5, the decoding device 3 acquires an intra prediction mode from the stream output from the encoding device 1 in step S201.

In step S<b>202 , the decoding device 3 determines whether or not the CU is divided into a plurality of TUs based on flag information included in the stream output from the encoding device 1 . If it is determined in step S202 that the CU is divided into multiple TUs, the operation proceeds to step S203. On the other hand, in step S202, CU
is not divided into multiple TUs, the operation proceeds to step S205.

In step S205, the decoding device 3 performs predetermined prediction on the TU to be decoded using the decoded reference pixels adjacent to the left and upper sides of the TU.

In step S203, the decoding device 3 determines whether the direction of the intra prediction mode is the direction from the lower left to the upper right or the direction from the upper right to the lower left. When it is determined in step S203 that the direction of the intra prediction mode is the direction from the lower left to the upper right or the direction from the upper right to the lower left, the operation proceeds to step S206.

On the other hand, when it is determined in step S203 that the direction of the intra prediction mode is other than the direction from the lower left to the upper right and the direction from the upper right to the lower left, the operation proceeds to step S204.

In step S204, the decoding device 3 adopts the conventional raster scan order (Z type as shown in FIG. 8A) used in HEVC as the decoding order described above.

When it is determined that the direction of the intra prediction mode is the direction from the lower left to the upper right (step S206), in step S207, the decoding device 3 sets the above-described decoding order as follows:
TU#A3 (bottom left TU in CU#A)→TU#A4 (bottom right TU in CU#A)→TU
#A1 (upper left TU in CU#A) → TU#A2 (upper right TU in CU#A) in decoding order, or TU#A3 (lower left TU in CU#A) → TU#A1 ( Upper left TU in CU#A)→TU#A4 (Lower right TU in CU#A)→TU#A2 (Upper right TU in CU#A
), a predetermined decoding order is adopted.

On the other hand, when it is determined that the direction of the intra prediction mode is not the direction from the lower left to the upper right (step S206), in step S211, the decoding device 3 sets TU#A2 (within CU#A) as the decoding order described above. upper right TU) → TU#A4 (lower right TU in CU#A)
→ TU#A1 (upper left TU in CU#A) → TU#A3 (lower left TU in CU#A), or TU#A2 (upper right TU in CU#A) → TU# Predefined decoding in the decoding order of A1 (upper left TU in CU#A) → TU#A4 (lower right TU in CU#A) → TU#A3 (lower left TU in CU#A) Adopt order.

In step S208, the decoding device 3 determines whether or not the reference pixel adjacent to the upper side of the TU to be decoded has been decoded. In step S208, if the decoding has been completed, the operation proceeds to step S209; otherwise, the operation proceeds to step S209.
Proceed to 210.

In step S209, the decoding device 3 performs predetermined prediction on the TU to be decoded using the decoded reference pixels adjacent to the left side, upper side, and lower side of the TU.

In step S210, the decoding device 3 performs predetermined prediction on a TU to be decoded using decoded reference pixels adjacent to the left side and lower side of the TU.

In step S212, the decoding device 3 determines whether or not the reference pixel adjacent to the left side of the TU to be decoded has been decoded. In step S212, if the decoding has been completed, the operation proceeds to step S213, and if the decoding has not been completed, the operation proceeds to step S213.
Proceed to 214.

In step S213, the decoding device 3 performs predetermined prediction on the TU to be decoded using the decoded reference pixels adjacent to the left, upper and right sides of the TU.

In step S214, the decoding device 3 performs predetermined prediction on the TU to be decoded using the decoded reference pixels adjacent to the right and upper sides of the TU.

In step S215, the decoding device 3 vertically inverts the signal obtained by the entropy decoding process and the inverse quantization process, and performs subsequent processes.

In step S216, the decoding device 3 horizontally inverts the signal obtained by the entropy decoding process and the inverse quantization process, and performs subsequent processes.

In step S217, the decoding device 3 vertically and horizontally inverts the signal obtained by the entropy decoding process and the inverse quantization process, and then performs subsequent processes.

According to the decoding device 3 according to the present embodiment, when at least one of the reference pixels on the lower side and the right side is used in intra prediction and the TransformSkip mode is applied (obtained by entropy decoding processing and inverse quantization processing) Even if the inverse orthogonal transform processing is not applied to the signal obtained by the conversion, it is possible to suppress the deterioration of the coding efficiency.

(Other embodiments)
As noted above, the present invention has been described through the above-described embodiments, but the statements and drawings forming part of the disclosure in such embodiments should not be understood to limit the present invention. From such disclosure, various alternative embodiments, implementations and operational techniques will become apparent to those skilled in the art.

Moreover, although not particularly mentioned in the above-described embodiment, a program may be provided that causes a computer to execute each process performed by the above-described encoding device 1 and decoding device 3 . Also, such a program may be recorded on a computer-readable medium. Such programs can be installed on a computer using a computer readable medium. Here, the computer-readable medium on which the program is recorded may be a non-transitory recording medium. The non-transitory recording medium is not particularly limited, but may be, for example, a recording medium such as a CD-ROM or a DVD-ROM.

Alternatively, a chip configured by a memory storing a program for realizing at least part of the functions in the encoding device 1 and the decoding device 3 described above and a processor executing the program stored in the memory may be provided. good.

1 Encoding device 11 Intra prediction mode determination unit 12 TU division determination unit 13 Encoding order control unit 14 Sequential local decoded image generation unit 14a Intra prediction unit 14b Residual signal generation unit 14c Orthogonal transformation/ Quantization unit 14d: Inverse quantization unit/inverse orthogonal transformation unit 14e: Local decoded image generation unit 15: Memory 16: Entropy encoding unit 3: Decoding device 31: Entropy decoding unit 32: Decoding order control unit 33: Sequential local decoding Image generating unit 33a...Intra prediction unit 33b...Inverse quantization/inverse transform unit 33c...Decoded image generating unit 34...Memory

Claims (4)

An encoding device that encodes a frame-based original image that constitutes a moving image by dividing it into blocks,
An intra prediction unit that generates a predicted image of the target block using the intra prediction mode;
a residual signal generation unit that generates a residual signal from the difference between the predicted image and the original image,
When orthogonal transform processing is not applied , the residual signal generation unit
horizontally inverting the residual signal when the predicted image is generated using at least the reference pixels positioned to the left and the lower side of the target block;
vertically inverting the residual signal when the predicted image is generated using at least reference pixels positioned to the right and upper sides of the target block ;
inverting the residual signal in the vertical and horizontal directions when the predicted image is generated using the reference pixels located on the left side and the upper side of the target block;
The residual signal is not inverted when the predicted image is not generated using the reference pixels located on either the left side or the upper side of the target block.
An encoding device characterized by: A decoding device that decodes an encoded stream by dividing an original image of each frame that constitutes a moving image into blocks,
An intra prediction unit that generates a predicted image of the target block using the intra prediction mode;
An inverse transform unit that performs inverse transform processing on the signal obtained by the entropy decoding process and the inverse quantization process ,
When the inverse transform processing is not applied , the inverse transform unit
horizontally inverting the signal when the predicted image is generated using at least reference pixels on the left side and the lower side of the target block;
when the predicted image is generated using at least the reference pixels on the right side and the upper side of the block to be encoded, inverting the signal in the vertical direction;
if the predicted image is generated using reference pixels positioned to the left and upper sides of the target block, inverting the signal in the vertical and horizontal directions;
The signal is not inverted when the predicted image is not generated using the reference pixels located on either the left side or the upper side of the target block.
A decoding device characterized by: A program for causing a computer to function as the encoding device according to claim 1. A program for causing a computer to function as the decoding device according to claim 2.

Images