WO2013065259A1

WO2013065259A1 - Video encoding device, video decoding device, video encoding method, video decoding method, and program

Info

Publication number: WO2013065259A1
Application number: PCT/JP2012/006812
Authority: WO
Inventors: 慶一蝶野; 啓史青木
Original assignee: 日本電気株式会社
Priority date: 2011-11-01
Filing date: 2012-10-24
Publication date: 2013-05-10

Abstract

This video encoding device is provided with a non-square frequency-transformation means and a non-square-scaling-matrix encoding means. The non-square-scaling-matrix encoding means predictively encodes the elements of a non-square scaling matrix using the elements of an already-encoded non-square scaling matrix. This video decoding device is provided with a non-square inverse frequency-transformation means and a non-square-scaling-matrix decoding means. The non-square-scaling-matrix decoding means predictively decodes the elements of a non-square scaling matrix using the elements of an already-decoded non-square scaling matrix.

Description

Video encoding apparatus, video decoding apparatus, video encoding method, video decoding method, and program

The present invention relates to a video encoding device and a video decoding device that use a scaling matrix in quantization processing and inverse quantization processing for a non-square quadtree frequency transform block.

Non-Patent Document 1 discloses a video encoding method using a scaling matrix in quantization processing and inverse quantization processing for a square frequency transform block.

Non-Patent Document 1 supports 4 × 4 and 8 × 8 square frequency conversion blocks. Therefore, a scaling matrix can be set for each of intra prediction / inter prediction and luminance (Y) and color difference (U and V) components. That is, six 4 × 4 scaling matrices (two combinations of intra prediction / inter prediction and three signal components Y / U / V) and two 8 × 8 scaling matrices (two intra prediction / inter prediction and signal components) One combination of Y) can be set. Further, the six 4 × 4 scaling matrices and the two 8 × 8 scaling matrices can be transmitted using Sequence Parameter Set (SPS) and Picture Parameter Set (SPS) (for details, refer to 7.3.2.1.1 of Non-Patent Document 1). (Listed in Sequence, parameter, set, data, syntax, and 7.3.2.2, Picture, parameter, set, RBSP, syntax).

7.3.2.1.1.1 １Scaling list syntax of Non-Patent Document 1 describes a method for predictively encoding elements of a scaling matrix based on a zigzag scan in order to reduce the code amount of each square scaling matrix in SPS and PPS. ing. That is, predictive coding closed in the scaling matrix is used. Hereinafter, this is referred to as intra-scaling matrix predictive coding.

1 includes a predictor 101, a frequency converter 102, a quantizer 103, an entropy encoder 104, an inverse quantization / inverse frequency converter 105, a buffer 106, and intra-scaling matrix prediction. The encoder 107 is configured.

In the video encoding device of the prior art, the residual signal (residual block signal) obtained by subtracting the prediction signal (predicted block signal) generated by the predictor 101 から from the input signal (input block signal) is output by the frequency converter 102. Obtain the residual frequency transform coefficient.

Furthermore, the quantizer 103 quantizes the residual frequency transform coefficient using an externally set scaling matrix to obtain a residual level.

The entropy encoder 104 performs entropy encoding on the residual level and outputs a bit stream.

For the subsequent prediction of the input signal, the residual level is converted into a reconstructed residual signal (reconstructed residual block signal) via the inverse quantization / inverse frequency converter 105, and reconstructed by adding the prediction signal. The buffer 106 is stored as a signal (reconstructed block signal).

The intra-scaling matrix prediction encoder 107 performs intra-scaling matrix prediction encoding. The intra-scaling matrix prediction encoder 107 supplies the prediction error of each element of the scaling matrix to the entropy encoder 104.

By the way, Non-Patent Document 2 describes a video code that uses non-square frequency transform blocks such as 2 × 8, 8 × 2, 4 × 16, 16 × 4, 8 × 32, and 32 × 8 in addition to square frequency transform blocks. Is disclosed. In Patent Document 2 that supports quadtrees, non-square frequency conversion is called non-square quadtree frequency conversion.

Considering the combination of Non-Patent Document 1 and Non-Patent Document 2, similar to the square frequency transform block, the non-square frequency transform block also includes intra prediction / inter prediction and luminance (Y) component and color difference (U and V). It is conceivable to set a scaling matrix for each component.

That is, in addition to the six 4 × 4 scaling matrices and the two 8 × 8 scaling matrices, six 2 × 8 scaling matrices (three combinations of intra prediction / inter prediction and signal component Y / U / V) And six 8 × 2 scaling matrices (two combinations of intra prediction / inter prediction and signal component Y / U / V) can be set. Similarly, six 4 × 16, 16 × 4, 8 × 32, and 32 × 8 scaling matrices can be set.

Figure 2 shows the frequency components of 2x8 and 8x2 non-square frequency transform blocks and their transmission order based on zigzag scan. FIG. 3 shows a state of intra-scaling matrix prediction encoding of 2 × 8 (2 rows and 8 columns) and 8 × 2 (8 rows and 2 columns) non-square scaling matrices. As can be seen from FIG. 3, as long as the intra-scaling matrix prediction encoding is used, the code amount of the scaling matrix increases by the number of additional 2 × 8 and 8 × 2 non-square scaling matrices.

An object of the present invention is to more suitably reduce the code amount of a non-square scaling matrix by non-square scaling matrix inter-prediction encoding.

A video encoding device according to the present invention is a video encoding device including non-square frequency conversion means and non-square scaling matrix encoding means, wherein the non-square scaling matrix encoding means converts the elements of the non-square scaling matrix. Predictive encoding is performed using elements of an encoded non-square scaling matrix.

A video decoding device according to the present invention is a video decoding device comprising non-square inverse frequency conversion means and non-square scaling matrix decoding means, wherein the non-square scaling matrix decoding means has already decoded the elements of the non-square scaling matrix. Predictive decoding is performed using elements of a non-square scaling matrix.

A video encoding method according to the present invention is a video encoding method executed by a video encoding device including non-square frequency conversion means and non-square scaling matrix encoding means, wherein the non-square scaling matrix encoding means includes The non-square scaling matrix elements are predictively encoded with the encoded non-square scaling matrix elements.

A video decoding method according to the present invention is a video decoding method executed by a video decoding device including non-square inverse frequency conversion means and non-square scaling matrix decoding means, wherein the non-square scaling matrix decoding means includes non-square scaling. It is characterized by predictively decoding the elements of the matrix with the elements of the decoded non-square scaling matrix.

A video encoding program according to the present invention provides a computer that implements a non-square frequency conversion unit and a non-square scaling matrix encoding unit, and predicts a non-square scaling matrix element using an encoded non-square scaling matrix element. It is characterized in that the processing to be converted is executed.

The video decoding program according to the present invention is a computer that realizes non-square inverse frequency conversion means and non-square scaling matrix decoding means, and performs processing for predictive decoding of non-square scaling matrix elements using decoded non-square scaling matrix elements. Is executed.

The code amount of the non-square scaling matrix can be more suitably reduced by the non-square scaling matrix inter-prediction encoding.

It is a block diagram of the video coding apparatus of a prior art. It is explanatory drawing which shows the frequency component (a number is the transmission order of each frequency component based on a zigzag scan) of a 2 * 8 and 8 * 2 non-square frequency conversion block. It is explanatory drawing which shows the mode of the prediction encoding in a scaling matrix of each of 2 * 8 and 8 * 2 non-square scaling matrix. It is explanatory drawing which shows a mode that an 8x2 non-square scaling matrix is predictively encoded using the transmitted 2x8 non-square scaling matrix. It is a block diagram of the video coding apparatus of 1st Embodiment. It is the flowchart which paid its attention to operation | movement of the non-square scaling matrix prediction encoder. It is a block diagram of the video decoding apparatus of 2nd Embodiment. It is the flowchart which paid its attention to operation | movement of the non-square scaling matrix prediction decoder 206. It is a block diagram which shows the structural example of the information processing apparatus using this invention.

The present invention takes advantage of the correlation between non-square scaling matrices of the same size, such as an additional 2 × 8 and 8 × 2, to encode the non-square scaling matrix to be transmitted first in the scaling matrix and subsequently transmit it The non-square scaling matrix is predictively encoded with the transmitted non-square scaling matrix. Hereinafter, it is referred to as non-square scaling matrix predictive coding. FIG. 4 shows how an 8 × 2 non-square scaling matrix is predictively encoded using the transmitted 2 × 8 non-square scaling matrix. With the non-square scaling matrix inter-prediction encoding, the code amount of the non-square scaling matrix can be more suitably reduced.

Embodiment 1. FIG.
The video encoding apparatus of the present embodiment shown in FIG. 5 includes a predictor 101, a frequency converter 102, a quantizer 103, an entropy encoder 104, an inverse quantization / inverse frequency converter 105, a buffer 106, and a scaling matrix. A predictive encoder 107 and a non-square scaling inter-matrix predictive encoder 108 are provided. The video encoding device predictively encodes each corresponding frequency component (by transposition) between non-square scaling matrices.

In the video encoding apparatus of the present embodiment, the residual signal (residual block signal) obtained by subtracting the prediction signal (predicted block signal) generated by the predictor 101 from the input signal (input block signal) is converted into the frequency converter 102. Is converted into a residual frequency conversion coefficient. The frequency conversion is a square frequency conversion and a non-square frequency conversion. The residual frequency transform coefficient obtained by the square frequency transform is called a square residual frequency transform coefficient. The residual frequency transform coefficient obtained by the non-square frequency transform is called a non-square residual frequency transform coefficient.

Furthermore, the quantizer 103 quantizes the residual frequency transform coefficient using an externally set scaling matrix to obtain a residual level. For a square residual frequency transform coefficient, a square residual level is obtained using a square scaling matrix. For non-square residual frequency transform coefficients, non-square residual levels are obtained using a non-square scaling matrix.

For the subsequent prediction of the input signal, the residual level is converted into a reconstructed residual signal (reconstructed residual block signal) via the inverse quantization / inverse frequency converter 105, and reconstructed by adding the prediction signal. The buffer 106 is stored as a signal (reconstructed block signal). However, the inverse quantization / inverse frequency converter 105 uses a square scaling matrix for inverse quantization of the square residual level. A non-square scaling matrix is used for inverse quantization of non-square residual levels.

The intra-scaling-matrix prediction encoder 107 encodes the square scaling matrix within the scaling matrix. Further, the horizontally long non-square scaling matrix (non-square scaling matrix 1) is subjected to intra-scaling matrix predictive coding. For example, in the case of the 2 × 8 and 8 × 2 non-square scaling matrix shown in FIG. 2, the non-square scaling matrix (a) is subjected to intra-scaling prediction encoding.

For example, if a 2 × 8 non-square scaling matrix (non-square scaling matrix 1) for an intra prediction block is represented by the following pseudo code:
Quant_intra_2 × 8 [2] [8] =
{
{6,10,13,16,18,23,25,27},
{18,23,25,27,29,31,33,36},
};
Prediction error from previous element in zigzag scan order in non-square scaling matrix 1, 6 (= 6-0), 12 (= 18-6), -8 = (10-18),…, 9 (= 36-27) are supplied to the entropy encoder 104.

Similarly, if the 2 × 8 non-square scaling matrix (non-square scaling matrix 1) for the inter prediction block is represented by the following pseudo code:

Quant_inter_2 × 8 [2] [8] =
{
{9,13,15,17,19,21,22,24},
{19,21,22,24,25,27,28,30}
};
Prediction error from previous element in zigzag scan order in non-square scaling matrix 1, 9 (= 9-0), 10 (= 19-9), -6 = (13-19),…, 6 (= 30-24) is supplied to the entropy encoder 104.

Subsequently, the non-square scaling matrix prediction encoder 108 する performs a non-square scaling matrix prediction encoding on the vertically long non-square scaling matrix.

For example, if an 8 × 2 non-square scaling matrix (non-square scaling matrix 2) for an intra prediction block is represented by the following pseudo code:
Quant_intra_8 × 2 [8] [2] =
{
{6,18},
{10,23},
{13,25},
{16,27},
{18,29},
{23,31},
{25,33},
{27,36}
};
Prediction error with non-square scaling matrix 1 transpose matrix elements, 0 (= 6-6), 0 (= 18-18), 0 = (10-10), ..., 0 (= 36-36) entropy This is supplied to the encoder 104.

Similarly, if the 2 × 8 non-square scaling matrix (non-square scaling matrix 2) for the inter prediction block is expressed by the following pseudo code:
Quant_inter_8 × 2 [8] [2] =
{
{9,19},
{13,21},
{15,22},
{17,24},
{19,25},
{21,27},
{22,28},
{24,30},
};
Prediction error with non-square scaling matrix 1 transpose matrix elements, 0 (= 9-9), 0 (= 19-19), 0 = (13-13),…, 0 (= 30-30) entropy This is supplied to the encoder 104.

Here, for each of intra prediction and inter prediction, a matrix obtained by transposing a 2 × 8 non-square scaling matrix (non-square scaling matrix 1) matches an 8 × 2 non-square scaling matrix (non-square scaling matrix 2). Note that the prediction errors are all zero.

FIG. 6 is a flowchart focusing on the operation of the non-square scaling inter-matrix predictive encoder 108, which is a feature of the present invention.

The non-square scaling matrix inter-prediction encoder 108 予測 transposes the transmitted non-square scaling matrix 1 in step S101.

Further, in step S102, the non-square scaling matrix inter-prediction encoder 108 calculates the difference of each element according to the zigzag scan with respect to the transposed matrix generated in step S101 and the non-square scaling matrix 2.

Embodiment 2. FIG.
The video decoding apparatus according to the present embodiment shown in FIG. 7 includes an entropy decoder 201, an inverse quantization / inverse frequency converter 202, a predictor 203, a buffer 204, an intra-scaling prediction decoder 205, and a non-square scaling matrix prediction. A decoder 206 is provided.

The entropy decoder 201 entropy-decodes the bitstream, and entropy-decodes the prediction error and residual level of each element of each scaling matrix. The prediction errors of the square scaling matrix and the horizontally long non-square scaling matrix (non-square scaling matrix 1) are supplied to the intra-scaling matrix prediction decoder 205. The prediction error of the vertically long non-square scaling matrix (non-square scaling matrix 2) is supplied to a non-square scaling matrix prediction decoder 206.

The intra-scaling matrix prediction decoder 205 predictively decodes the prediction errors of the square scaling matrix and the horizontally long non-square scaling matrix (non-square scaling matrix 1).

The intra-scaling matrix prediction decoder 205 supplies the square scaling matrix and the non-square scaling matrix (non-square scaling matrix 1) subjected to the intra-scaling matrix prediction decoding to the inverse quantization / inverse frequency converter 202. The non-square scaling matrix (non-square scaling matrix 1) is supplied to the non-square scaling matrix inter-prediction decoder 206.

The non-square scaling matrix prediction decoder 206 predicts and decodes the prediction error of the vertically long non-square scaling matrix (non-square scaling matrix 2). For example, when a 2 × 8 non-square scaling matrix (non-square scaling matrix 1) for an intra prediction block decoded by the scaling matrix intra prediction decoder 205 is expressed by the following pseudo code:
Quant_intra_2 × 8 [2] [8] =
{
{6,10,13,16,18,23,25,27},
{18,23,25,27,29,31,33,36},
};
Add 8x2 non-square scaling matrix (non-square scaling matrix 2) zigzag scan order prediction error, 0, 0, 0, ..., 0 to zigzag scan order element of non-square scaling matrix 1 transpose matrix The non-square scaling matrix 2
Quant_intra_8 × 2 [8] [2] =
{
{6,18},
{10,23},
{13,25},
{16,27},
{18,29},
{23,31},
{25,33},
{27,36}
};
Calculate

Similarly, the intra-scaling matrix prediction decoder 205 expresses a 2 × 8 non-square scaling matrix (non-square scaling matrix 1) for the decoded inter prediction block by the following pseudo code:
Quant_inter_2 × 8 [2] [8] =
{
{9,13,15,17,19,21,22,24},
{19,21,22,24,25,27,28,30}
};
Add 8x2 non-square scaling matrix (non-square scaling matrix 2) zigzag scan order prediction error, 0, 0, 0, ..., 0 to zigzag scan order element of non-square scaling matrix 1 transpose matrix The non-square scaling matrix 2
Quant_inter_8 × 2 [8] [2] =
{
{9,19},
{13,21},
{15,22},
{17,24},
{19,25},
{21,27},
{22,28},
{24,30},
};
Calculate

The inverse quantization / inverse frequency converter 202 performs inverse quantization / inverse frequency conversion on the supplied residual level and outputs a reconstructed residual signal. The reconstructed residual signal is added with the prediction signal supplied from the predictor 203 and stored in the buffer 204 as a reconstructed signal.

However, the inverse quantization / inverse frequency converter 202 calculates the reconstructed residual signal using the square scaling matrix for the square residual level. For non-residual levels, a reconstructed residual signal is calculated using a non-square scaling matrix.

FIG. 8 is a flowchart focusing on the operation of the non-square scaling inter-matrix prediction decoder 206, which is a feature of the present invention.

The non-square scaling matrix inter-prediction decoder 206 transposes the decoded non-square scaling matrix 1 in step S201.

In step S202, each element of the non-square scaling matrix 2 is calculated according to the zigzag scan based on the prediction error of the non-square scaling matrix 2 and the transposed matrix generated in step S201.

This is the end of the description of the operation of the video decoding apparatus according to the present embodiment.

Embodiment 3. FIG.
In the above-described embodiment, encoding and decoding of 2 × 8 and 8 × 2 non-square scaling matrices are shown. Similar encoding and decoding can naturally be applied to non-square scaling matrices of 4 × 16 and 16 × 4 and 8 × 32 and 32 × 8.

In the above-described embodiment, the horizontally long non-square scaling matrix is predicted within the scaling matrix, and the vertically long non-square scaling matrix is predicted between the non-square scaling matrices. A vertically long non-square scaling matrix may be predicted within a scaling matrix, and a horizontally long non-square scaling matrix may be predicted between non-square scaling matrices.

In the embodiment described above, the non-square scaling matrix 2 to be encoded / decoded subsequently is always predicted between non-square scaling matrices. When the correlation between non-square scaling matrices is low, intra-scaling matrix prediction may be applied to the non-square scaling matrix 2 as well. In such an embodiment, the video encoding device may embed auxiliary information on whether non-square scaling matrix 2 or intra-scaling matrix prediction or non-square scaling matrix prediction is applied to the bit stream. . The video decoding device may perform decoding based on intra-scaling matrix prediction or non-square scaling matrix prediction on the non-square scaling matrix 2 based on the auxiliary information.

In the embodiment described above, the non-square scaling matrix 2 to be encoded / decoded subsequently is predicted between non-square scaling matrices, and all the prediction errors of the elements are transmitted. When the transposition of the non-square scaling matrix 1 and the non-square scaling matrix 2 completely match, instead of transmitting a prediction error, the video encoding apparatus transposes the non-square scaling matrix 1 and the non-square scaling matrix 2 It is only necessary to embed auxiliary information indicating that the data is in the bitstream. The video decoding device may use the non-square scaling matrix 2 as a result of transposing (adding 0 to) the non-square scaling matrix 1 based on the auxiliary information.

Furthermore, in the embodiment of the invention described above, as is apparent from the above description, it can be configured by hardware, but can also be realized by a computer program.

The information processing system shown in FIG. 9 includes a processor 1001, a program memory 1002, a storage medium 1003 for storing video data, and a storage medium 1004 for storing a bitstream. The storage medium 1003 and the storage medium 1004 may be separate storage media, or may be storage areas composed of the same storage medium. A magnetic storage medium such as a hard disk can be used as the storage medium.

In the information processing system shown in FIG. 9, the program memory 1002 stores a program for realizing the function of each block shown in FIG. 5 and FIG. The processor 1001 implements the functions of the video encoding device or the video decoding device shown in FIGS. 5 and 7 by executing processing according to the program stored in the program memory 1002.

As mentioned above, although this invention was demonstrated with reference to embodiment and an Example, this invention is not limited to the said embodiment and Example. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

This application claims priority based on Japanese Patent Application No. 2011-240539 filed on November 1, 2011, the entire disclosure of which is incorporated herein.

The present invention can be suitably applied to a video encoding device, a video decoding device, or the like that uses a scaling matrix in quantization processing and inverse quantization processing for a non-square quadtree frequency transform block.

101 Predictor 102 Frequency converter 103 Quantizer 104 Entropy encoder 105 Inverse quantization / inverse frequency converter 106 Buffer 107 Intra-scaling matrix prediction encoder 108 Non-square scaling inter-matrix prediction encoder 201 Entropy decoder 202 Inverse quantization / inverse frequency converter 203 Predictor 204 Buffer 205 Intra-scaling matrix prediction decoder 206 Non-square scaling inter-matrix prediction decoder

Claims

A video encoding device comprising non-square frequency conversion means and non-square scaling matrix encoding means,
The non-square scaling matrix encoding means predictively encodes the elements of the non-square scaling matrix with the elements of the encoded non-square scaling matrix.
The video encoding device according to claim 1, wherein the non-square scaling matrix encoding unit calculates a difference between each element for a non-square scaling matrix and a transposed matrix of the encoded non-square scaling matrix.
A video decoding device comprising non-square inverse frequency conversion means and non-square scaling matrix decoding means,
The non-square scaling matrix decoding means predictively decodes the elements of the non-square scaling matrix with the elements of the decoded non-square scaling matrix.
The video decoding device according to claim 3, wherein the non-square scaling matrix decoding means calculates elements of the non-square scaling matrix based on a transposed matrix and a non-square scaling matrix of the decoded non-square scaling matrix.
A video encoding method executed by a video encoding device including non-square frequency conversion means and non-square scaling matrix encoding means,
The video encoding method, wherein the non-square scaling matrix encoding means predictively encodes the elements of the non-square scaling matrix with the elements of the encoded non-square scaling matrix.
A video decoding method executed by a video decoding device comprising non-square inverse frequency conversion means and non-square scaling matrix decoding means,
The video decoding method, wherein the non-square scaling matrix decoding means predictively decodes the elements of the non-square scaling matrix with the elements of the decoded non-square scaling matrix.
A computer that implements non-square frequency conversion means and non-square scaling matrix encoding means,
A video encoding program for executing a process of predictively encoding non-square scaling matrix elements with encoded non-square scaling matrix elements.
A computer that implements non-square inverse frequency conversion means and non-square scaling matrix decoding means,
A video decoding program for executing a process of predictive decoding an element of a non-square scaling matrix with an element of a decoded non-square scaling matrix.