JPH09284764A - Video signal encoding method and video signal encoder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the compression efficiency of data by encoding video frame signals including an object by an extension interpolation method. SOLUTION: A binary map detector 110 detects and encodes a binary map inside the video frame signals from a frame memory 50 by normal encoding, generates binary map signals provided with binary picture elements equal to the picture element number of the video frame signals and supplies them through a line LIO to an E/I device 400. In the meantime, the signals from the memory 50 are divided into plural processing blocks in a block generator 200 and supplied to a switching circuit 300 controlled by control signals CS for indicating whether or not an object boundary part inside a video frame is present in the respetive processing blocks by a block unit. When the boundary part is present, supply is performed to the device 400, an extended processing block is obtained and further, it is rescaled and encoded in the conversion encoder 510 of a second code channel 500. Thus, the device 400 can improve the data compression efficiency in the channel 500.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for encoding a video signal at a low bit rate,
In particular, the present invention relates to a method and apparatus for encoding a video signal using an extension-interpolation (EI) method.

[0002]

As is well known, video signals may be transmitted in digital form in a variety of electronic applications such as high definition television (HDTV) systems or video telephone systems. A series of video "frames"
When the video signal consisting of
A large amount of digital data is generated because each line of the video frame is defined by a series of digital data elements called "pixels". However, due to the limited frequency bandwidth available on conventional transmission channels, the amount of data needs to be compressed or reduced for transmitting large amounts of digital data through the limited channel.

One of the methods for encoding a video signal in a low bit rate video signal encoding system is as follows.
This is a so-called object-oriented analysis-synthesis coding method (Mi.
chael Hotter's paper, "Object-O
Oriented Analysis-Synthesi
s Coding Based on MovingT
wo-Dimensional Objects ", S
Signal Processing: Image Co
mmunication, 2 pp. 409-428 (19
90))).

According to this object-oriented analysis-synthesis coding method, an input video signal having a moving object is divided into a plurality of objects, and three sets of parameters that define the movement, contour and pixel data of each object. Are processed through different channels.

In object-oriented analysis-synthesis coding techniques,
When processing image data or pixels in an object, transform coding techniques are mainly used to reduce the spatial redundancy contained in the image data. One of the transform coding techniques mainly used for compression of video data is DC in block units.
T (discrete cosine transform). This DCT transforms a block of digital video data, for example, a block of 8 × 8 pixels, into a set of transform coefficient data. This encoding method is described in, for example, the article by Chen and Pratt,
"Scene Adaptive Coder", IE
EE Transactions on Commun
ications, COM-32, No. 3,225-
232 (March 1984). DCT
Although it is not used as frequently as, the DST (Discrete Sine Transform), the Hartley transfor
m) or other transform methods may be used for block transform coding.

In the block-by-block encoding, the area other than the background or the object in the block is transformed after being filled with, for example, 0, the average value of the pixels in the object area in the block, or the mirror image. Referring to FIGS. 2 and 3, a conventional method for filling a background area is illustrated as one-dimensional data. More specifically, in FIG. 2, the background area is filled with zero values, and in FIG. 3, the background area is filled with the average value of the pixel values of the object area.

Such an encoding method is an ordinary encoding method (for example, Joint Photographic E
xperts Group; JPEG, Moving
pictures Experts Group; MP
Even if the two-dimensional DCT block used in EG, H261) can be used, it may be insufficient from the viewpoint of data compression efficiency because unnecessary or unwanted data may be included in the background area of the image. There is a point.

[0008]

SUMMARY OF THE INVENTION Therefore, it is an object of the present invention to effectively encode a video frame signal containing an object by using an extension-interpolation technique, and a video signal encoding with a better data compression efficiency. Is to provide a method.

Another object of the present invention is to provide a video signal coding apparatus for coding a video frame signal containing an object using a block-by-block coding method and an extension-interpolation technique.

[0010]

In order to achieve the above-mentioned object, according to the present invention, it is composed of N × N pixels (N is a positive integer) included in a video frame signal having an object. A video signal encoding method for converting a processing block to encode a video signal, wherein the pixels include an object pixel representing a pixel located inside an object and a background pixel representing a pixel located outside the object. A first step of selecting L columns of the processing block having at least one object pixel (L is an integer from 0 to N), and for each column selected from the first step. To determine M representing the number of object pixels included in each column, and to supply a first vector in which each element is a pixel value of each of the M object pixels (M
Is an integer from 1 to N), and for each column, one extension matrix is selected from a plurality of predetermined extension matrices based on the M value and the N value, and A third step of providing the expanded first vector by multiplying the first vector by the selected expansion matrix; and N pixels having a value for each element of the expanded first vector. A fourth step of providing an expanded first processing block having L columns, and for each row of the expanded first processing block,
A fifth step in which each element supplies an L-dimensional second vector that is a value of a pixel included in each row of the expanded first processing block, and is predetermined based on the L value and the N value. A sixth step of selecting one of the plurality of extended matrices selected, and multiplying each of the second matrix and the extended matrix selected in the sixth step to obtain N second extended vectors. Supply a vector, said expanded second
N having N pixels whose pixel value is each element of the vector
A seventh step of providing an extended second processing block having a number of rows, counting the number of the object pixels in the processing block and determining the number of the object pixels by the number of pixels in the processing block. An eighth step of determining a block scaling factor that is a divided number, and a ninth step of multiplying each pixel of the extended second processing block by the block scaling factor to generate the extended processing block. A video signal encoding method is provided, which comprises:

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 5A shows a block of a digital video signal having 8 × 8 pixels, and each pixel is represented by a square. This block is composed of an object area represented by hatched pixels and other background areas. Pixels shown with diagonal lines are called "object pixels", and the other pixels are called "background pixels". The object pixel is shown in FIG.
As shown in FIG. 5B and FIG. 5C, the extension-interpolation of the present invention.
n; E-I) method is used to extend to fill the entire block. At this time, as shown in the figure, horizontal expansion and vertical expansion are individually performed. These horizontal expansion or vertical expansion is performed before the expansion in the other direction, and the priority order thereof may be determined by the characteristics of the image. Horizontal expansion or vertical expansion is performed in units of columns or rows. Block is Nx
If it contains N pixels, then for each column or row, an M-dimensional vector is converted to an N-dimensional vector. Here, M is an integer of 1 or more and N or less, elements of the M-dimensional vector represent M object pixel values included in each column or row, and elements of the N-dimensional vector are N expanded pixels. Represents a value. For example, in the case of the third column in the block shown in FIG. 5A, the 5D vector is converted into an 8D vector horizontally and vertically extended as shown in FIG. 5B.

The transformed M obtained by applying the M-point one-dimensional DCT to the M-dimensional vector f _1.
The dimension vector F ₁ is expressed by the following equation (1) as follows.

[0014]

[Equation 17]

[0015] Here, f ₁ (n ₁₎ is n ₁ th element of _{f 1, F 1 (k 1} ) is k ₁ th element of F _1, n ₁ and k ₁ is 0
Each represents an integer from M1 to M-1.

B _ij is expressed by the following equation (2).

[0017]

(Equation 18)

Similarly, the M-dimensional vector f ₁ is the E of the present invention.
Is extended with -I method, when forming the N-dimensional vector f _2, 1-dimensional D N-point in N-dimensional vector f ₂
The transformed N-dimensional vector F ₂ obtained by applying CT is expressed by the following equation (3).

[0019]

[Equation 19]

Where f ₂ (n ₂ ) is the n ₂ -th element of f ₂ , F ₂ (k ₂ ) is the k ₂ -th element of F ₂ , and n ₂ and k ₂ are 0.
It is an integer not less than N-1.

Further, a _ij is defined by the following equation (4).

[0022]

(Equation 20)

In the E-I method of the present invention, the M-dimensional vector f ₁ is expanded to the N-dimensional vector f ₂ without generating additional frequency domain data. That is, it is defined by the following equation (5).

[0024]

(Equation 21)

Here, μ ₀ is a scaling factor for making the DC components of f ₁ and f ₂ equal, and is given as follows.

[0026]

(Equation 22)

When the above equation (5) is satisfied, f ₁ is converted into f ₂ without generating additional data in the frequency domain, so that the E-I process is optimal.

Using the above equations (1) and (3), f ₂
Is calculated from f _{1 as} follows.

[0029]

(Equation 23)

Or

[0031]

(Equation 24)

Here, A represents an N × N matrix composed of the components a _ij used in the above equation (7A), and B represents an N × M matrix composed of the components b _ij . Equation (7)
A) and equation (7B) are simplified as the following equation (8A) or (8B).

[0033]

(Equation 25)

Or

[0035]

(Equation 26)

Where C is an N × M matrix,
Equal to A ^- 1B.

Using the above relations, an object of arbitrary shape is extended to fill the NxN block without generating additional elements in the frequency domain.

On the contrary, the original data of FIG.
The block is restored from the expanded block shown in (C).

If N equals M, then C is the identity matrix. Therefore, the expansion process does not change the original vector f ₁ and can be omitted.

As shown in FIGS. 5A to 5C, the third to seventh columns of the block of FIG. 5A are the optimum EI of the present invention. Method or a linear interpolation method is used to expand in the horizontal direction to form blocks as shown in FIG. Similarly, the row of the horizontally expanded block in FIG. 5B is vertically expanded using the optimal E-I method or the linear interpolation method to obtain a block as shown in FIG. 5C. Become.

So far, the EI method of the present invention has been described only for the case of DCT. However, other conversions,
For example, a method such as DST (discrete sine transform), Hadamard transform, or Haar transform may be used instead. When this DST is used to encode an N × N block, the E-I method uses the DCT except that a _ij and b _ij in the following equation (9) become the following equation (9). Is the same as in. That is,

A _ij is

[0043]

[Equation 27]

Also, b _ij is

[0045]

[Equation 28]

Is as follows.

When the correlation of video data is not large in the spatial domain, for example, in the case of interframe coding in which the difference between two adjacent frames is coded, DST
It is known that the block transform coding using Ĥ will perform better than using the DCT.

In the method described above, the scaling factor μ ₀ was applied only to the DC component in the frequency domain. However, it may be effective to apply this scaling factor μ ₀ to other components as well. In this case, the equation (7A) is transformed into the following equation (11).

[0049]

(Equation 29)

In this case, the total amount of energy contained in the processing block will be significantly reduced during the expansion process. In order to suppress such a reduction in energy, each pixel value of the expanded processing block is multiplied by the block scaling factor. The block scaling factor is defined as the number of object pixels in one processing block divided by N × N, which is the number of pixels in one block.

FIG. 6 shows a block diagram of a digital video signal coding apparatus according to the present invention. This coding device includes first and second coding channels 100 and 500.
And an extension-interpolation device (EI device) 400 for producing extended processing blocks for effectively encoding the boundary portion of the object in the video signal. The first coding channel 100 is used for coding the contour signal of the object, and the second coding channel 500 is used for coding the digital video signal block by block.

A digital video signal supplied from a known image source (not shown) (for example, a hard disk or a compact disk) is input and stored in the frame memory 50. The digital video signal of one frame includes an object, and includes an object pixel located inside the object and a background pixel located outside the object pixel. Here, the background pixel is represented as a pixel having a value that is significantly larger or smaller than the normal range of pixel values. Thereafter, the video frame signal fetched from the frame memory 50 is supplied to the binary map detector 110 of the first coding channel 100 and the block generator 200, respectively.

The first coding channel 100, which includes a binary map detector 110 and a binary map encoder 120, uses a conventional coding technique to map the binary map in the video frame signal from the frame memory 50. Detecting and encoding produces a coded binary map signal. This binary map has a number of binary pixels equal to the number of pixels of the video frame signal. Each binary pixel is determined by whether the corresponding pixel of the video frame signal is an object pixel or a background pixel. Then, the binary map detected by the binary map detector 110 is supplied to the binary map encoder 120 to be encoded and supplied to the E / I device 400 of the present invention through the line L10.

The binary map encoder 120 receives the binary map received from the binary map detector 110 from, for example,
It is encoded using the binary arithmetic code of JPEG and supplied to the formatting circuit 600.

On the other hand, the block generator 200 divides the video frame signal from the frame memory 50 into a plurality of processing blocks of the same size consisting of N × N pixels, and each processing block is a block-by-block switching circuit 300. Supply to. In this switching circuit 300, each processing block from the block generator 200 is selectively supplied to the E-1 device 400 or the second coding channel 500 in response to a control signal CS from a system controller (not shown). To do. The system controller generates the control signal CS based on the contour information of the object in the video frame signal. The control signal CS indicates whether or not the boundary portion of the object in the video frame exists in each processing block. If the boundary portion of the object exists within the processing block, that is, if the processing block has both an object region and a background region, the processing block is supplied to the E-I device 400 that generates the extended processing block,
If not, it is sent to the second encoded channel 500.

According to the present invention, the E-I device 400
Transforms each processing block from the switching circuit 300 into an extended processing block in order to improve the compression efficiency of the data in the second coding channel 500.
More specifically, the processing block as shown in FIG.
-Supplied to the I-device 400 and shown in FIGS.
Converted to extended processing blocks as described for (C). The detailed operation of the E / I device 400 will be described below with reference to FIG. 7.

The second coding channel 500, which includes a transform encoder 510, a quantizer 520 and an entropy encoder 530, is extended from the E-I device 400 by using conventional transform and statistical coding techniques. The video data included in each of the processed blocks or the unexpanded processing block from the switching circuit 300 is encoded. That is, the transform encoder 510 uses, for example, the discrete cosine transform to transform the image data in the spatial domain of each processing block supplied from the E-I device 400 or the switching circuit 300 into a set of transform coefficients in the frequency domain. While transforming, the quantizer 52
0. The quantizer 520 quantizes a set of input transform coefficients using a known quantizing method, and then supplies the quantized set of transform coefficients to the entropy encoder 530. Further processing.

The entropy encoder 530, for example,
Extended from quantizer 520 using a method that combines run-length and variable length coding techniques.
Alternatively, a quantized set of transform coefficients for each unexpanded processing block is encoded to generate an encoded video frame signal. Then, the video frame signal encoded by the entropy encoder 530 is supplied to the formatting circuit 600.

The formatting circuit 600 includes the encoded binary map from the binary map encoder 120 of the first encoding channel 100 and the second encoding channel 5.
00 coded video frame signal from entropy encoder 530 and provides the formatted digital video signal to a transmitter (not shown) for its transmission.

FIG. 7 shows a detailed block diagram of the expansion-interpolation device 400 according to the invention shown in FIG. The expansion-interpolation device 400 includes an object pixel counter 410, a start / size determination block 420, a control block 430, a RAM 440, a padding block 46.
0, extended matrix memory 470, extended block 48
0 and rescaling block 490.

The processing block from the switching circuit 300 of FIG. 6 is input to and stored in the RAM 440 through the line L20. The binary map from binary map detector 110 of FIG. 6 is input to object pixel counter 410 and start / size determination block 420 via line 110.

The object pixel counter 410 counts the number of object pixels in each processing block according to the binary map and supplies the object pixel to the rescaling block 490.

In the start / size determination block 420, the size and start signal are determined according to the binary map. The magnitude signal represents the number of object pixels in the currently processed column or row of the processing block, and the start signal represents the position of the first object pixel in the currently processed column or row.
The start signal and the magnitude signal are supplied to the padding block 460, and the magnitude signal is supplied to the expansion matrix memory 470 and the expansion block 480.

Detailed functions of the start / size determination block 420 will be described with reference to the examples of FIGS. 5A to 5C. If there are no object pixels in the current column, as in the first two columns in FIG. 5A, the magnitude signal is present in the current column for the padding block 460, the extended matrix memory 470 and the extended block 480. The extension-indicates that it does not need to be processed in the interpolation process. In the case of the third column of the processing block of FIG. 5A, the start signal indicates that the third pixel is the first object pixel, and the magnitude signal indicates that there are five object pixels in that column. Indicates.

During vertical expansion, the start signal and the magnitude signal are fixed at 3 and 6, respectively. This is because the starting position and the number of object pixels in each row of the horizontally expanded block are as shown in FIG.
This is because they are the same as shown by.

The extension matrix memory 470 stores an extension matrix for converting an M-dimensional vector into an N-dimensional vector, that is, C of the equation (8B). N is determined according to the system design, and is 8 in many cases. Therefore, the extension matrix C can be pre-calculated for all values of M, that is, 1 to N, and stored in the extension matrix memory 470. Therefore, the extension matrix memory 470 is composed of a ROM. You will get it.

Expansion-interpolation for a column (or row) of a block is done by expanding matrix memory 470 into an M-dimensional vector formed from the values of the object pixels of that column (or row).
This can easily be done by multiplying by the appropriate extension matrix stored in. The appropriate extension matrix is selected according to the magnitude signal and provided to the extension block 480 for performing the multiplication. If the magnitude signal indicates that there are no object pixels in the current column or row, then the expansion matrix is not selected and is not provided to expansion block 480.

On the other hand, in the control block 430, H
/ V and column / row number signals are generated. This H / V signal indicates whether the expansion-interpolation device 400 of the present invention is performing horizontal expansion or vertical expansion, and the column / row number signal is the expansion-interpolation device 40 of the present invention.
0 indicates the column or row currently processed. In general, the columns / rows of a block are processed sequentially, so these signals are determined by the system design, eg, the number of columns and rows of the processing block. H / V signal and column / row number signal are start /
It is supplied to the size determination block 420 and the RAM 440.

As described above, the switching circuit 300 of FIG. 8 supplies the processing block RAM 440.
The columns or rows of the processing block are H / H from the control block.
The signals are sequentially supplied from the RAM 440 according to the V signal and the column / row number signal. Illustratively, assume that the horizontal expansion is done first. In this case, each row of processing blocks is first supplied to padding block 460.

In the padding block 460, an M-dimensional vector representing the object pixel value included in the column supplied from the RAM 440 is formed in accordance with the start signal and the size signal and is supplied to the expansion block 480. This M-dimensional vector is generated by selecting M consecutive object pixels from the position represented by the start signal.

According to the magnitude signal, that is, the M-value signal, one of a plurality of predetermined expansion matrices stored in the expansion matrix memory 470 is selected and supplied to the expansion block 480. The extension matrix is multiplied with the M-dimensional vector at extension block 480 to form an extended M-dimensional vector. The N-dimensional vector or the horizontally expanded column is stored again in the RAM 440 at the position where the original column was stored.

As described above, one column of the processing blocks stored in the RAM 440 is converted into a horizontally expanded column and stored again. All columns in the processing block are
It is processed in a similar manner except for columns that do not contain object pixels. For columns that do not contain object pixels, the magnitude signal indicates that there are no object pixels in the extension block, so that no multiplication is performed and the original data in RAM 440 is maintained.

When the horizontal expansion is completed, the horizontally expanded processing block including the expanded column is stored in the RAM 440 as shown in FIG. 5B.

After the horizontal expansion is completed, the horizontally expanded rows of the processing blocks are sequentially input to the padding block 450 in the RAM 440 according to the H / V signal and the column / row number signal from the control block 430. Except for this, padding block 460 and extended matrix memory 47
0, the function of the expansion block 480 is the same as in the case of horizontal expansion. More specifically, the M-dimensional vector formed from each row of the horizontally expanded processing block is supplied from the padding block 450 to the expansion block 480, where:
Converted to N-dimensional vector. Expanded row (or N
The dimension vector) is stored in the RAM 440 at the original position of the corresponding row of the horizontally expanded processing block.

After horizontal expansion and vertical expansion are completed, RA
The expanded processing block stored in M440 is provided to rescaling block 490. In rescaling block 490, each pixel value of the extended processing block will be scaled down using a block scaling factor, which is supplied from object pixel counter 410. It is determined according to the number of object pixels per pixel.

The rescaled processing block is supplied to and encoded by the transform encoder 510 of the second encoding channel 500 of FIG.

While specific embodiments of the present invention have been described above, it will be apparent to those skilled in the art that various modifications can be made without departing from the scope of the claims set forth herein.

[0078]

Therefore, according to the present invention, by using the optimum expansion-interpolation method or the linear interpolation method to significantly reduce the frequency components generated between the pixels inside the object and the pixels outside the object, The coding efficiency can be increased.

[Brief description of drawings]

FIG. 1 is a diagram showing one-dimensional data including a background area and an object area.

FIG. 2 is a diagram showing a method of filling a background area with zero.

FIG. 3 is a diagram showing a method of filling a background area with an average value of an object area.

FIG. 4 is a diagram illustrating a method of filling a background area of the present invention.

FIG. 5 consists of (A), (B) and (C), and (A)
Is a video block having a background area and an object area, (B) is a horizontally expanded video block,
(C) is a schematic diagram showing a vertically expanded video block.

FIG. 6 is a block diagram illustrating a video signal encoding device of the present invention.

7 is a detailed block diagram of the extension-interpolation (EI) device of FIG.

[Explanation of symbols]

 50 frame memory 100 first coding channel 110 binary map detector 120 binary map encoder 200 block generator 300 switching circuit 400 E / I device 410 object pixel counter 420 start / size determination block 430 control block 440 RAM 460 padding block 470 extended matrix memory 480 extended block 490 rescaling block 500 second coding channel 510 transform encoder 520 quantizer 530 entropy encoder 600 formatting circuit CS control signal

Claims (5)

[Claims] 1. A video signal coding method for coding a video signal by converting a processing block composed of N × N pixels (N is a positive integer) included in a video frame signal having an object. The pixel is divided into an object pixel representing a pixel located inside the object and a background pixel representing a pixel located outside the object, and L of the processing blocks having at least one object pixel. Of columns (L is an integer of 0 or more and N or less), and M representing the number of object pixels included in each column is determined for each column selected from the first step. And a second step in which each element is a pixel value of each of the M object pixels (M is an integer of 1 or more and N or less); and for each column, the M value and Based on the N value,
A third step of selecting one extension matrix from a plurality of predetermined extension matrices and supplying the extended first vector by multiplying the selected extension matrix by the first vector, A fourth step of providing an extended first processing block having L columns with N pixels each having a value for each element of the extended first vector; A fifth step of supplying, for each row, an L-dimensional second vector in which each element is a value of a pixel included in each row of the expanded first processing block; Based on the sixth step of selecting one of a plurality of predetermined expansion matrix, based on each of the expansion matrix and the second vector selected in the sixth step, N expanded second
A seventh step of supplying a vector and supplying an expanded second processing block having N rows with N pixels having pixel values for each element of the expanded second vector; Counting the number of the object pixels in the processing block and determining a block scaling factor that is the number of the object pixels divided by the number of pixels in the processing block; and expanding the block scaling factor. And a ninth step of generating the extended processing block by multiplying each pixel of the generated second processing block. 2. The extension matrix for converting the M-dimensional vector into an N-dimensional extended vector is defined by the following equation: Here, a _ij is expressed as the following equation, and Also, b _ij is given by the following equation, and The above μ ₀ is as follows: The video signal encoding method according to claim 1, wherein: 3. The extension matrix for converting the M-dimensional vector into an N-dimensional extended vector is defined by the following equation: Here, b _ij is expressed by the following equation, and Also, a _ij is given by the following equation, and The above μ ₀ is as follows: The video signal encoding method according to claim 1, wherein: 4. A video signal encoding method for encoding a video signal by converting a processing block consisting of N × N pixels (N is a positive integer) included in a video frame signal having an object. The pixel is divided into an object pixel representing a pixel located inside the object and a background pixel representing a pixel located outside the object, and L of the processing blocks having at least one object pixel. Of columns (L is an integer of 0 or more and N or less), and M representing the number of object pixels included in each column is determined for each column selected from the first step. And a second step in which each element is a pixel value of each of the M object pixels (M is an integer of 1 or more and N or less); and for each column, the M value and Based on the N value,
A third step of selecting one extension matrix from a plurality of predetermined extension matrices and supplying the extended first vector by multiplying the selected extension matrix by the first vector, A fourth step of providing an extended first processing block having L columns with N pixels each having a value for each element of the extended first vector; A fifth step of supplying, for each row, an L-dimensional second vector in which each element is a value of a pixel included in each row of the expanded first processing block; Based on the sixth step of selecting one of a plurality of predetermined expansion matrix, based on each of the expansion matrix and the second vector selected in the sixth step, N expanded second
A seventh step of supplying a vector and supplying an expanded second processing block having N rows with N pixels having pixel values for each element of the expanded second vector; M-dimensional vector The extended matrix that transforms into an N-dimensional vector is expressed by the following equation: Where a _ij is expressed as In addition, b _ij is given by the following equation, and The above μ ₀ is as follows: A video signal encoding method characterized by: 5. A video signal coding for coding a video frame signal of a video having an object, the video frame signal comprising an object pixel representing a pixel located inside the object and a background pixel representing a pixel located outside the object. The apparatus is a binary map detecting means for detecting a binary map indicating which of the pixels included in the video frame signal is the object pixel, and the binary map is encoded and encoded. First encoding means for generating a first video frame signal; first dividing means for dividing the video frame into a plurality of processing blocks having N × N pixels (N is a positive integer); Control signal generation means for generating a control signal indicating whether or not a block includes both the object pixel and the background pixel; and, in accordance with the control signal, the processing block, the object pixel and the background pixel Second dividing means that divides the first processing block set having one side into a second processing block set other than the above, and each of the first processing block set is converted into an extended processing block. A second means encoded by encoding the conversion means and the set of the extended processing block or the second processing block;
A video signal coding apparatus comprising: a second coding means for supplying a video frame signal; and a formatting means for formatting the coded first and second video frame signals, wherein the converting means includes each column. Selects the processing block of L columns having at least one object pixel (L is an integer of 0 or more and N or less)
Selecting means, and for each column selected by the selecting means, determining M representing the number of object pixels included in each column, and
Each element supplies a first vector that is each value of the M object pixels included in each column (M is an integer of 0 or more and N or less)
A number determining means, and selecting one of the plurality of predetermined expansion matrices based on the L value and the N value,
Matrix selecting means for supplying an expanded first vector by multiplying the selected expanded matrix by a first vector, wherein the expanded matrix for converting an M-dimensional vector into an N-dimensional vector is as follows: Is defined as Here, a _ij is expressed by the following equation, and Also, b _ij is given by the following equation, and The above μ ₀ is as follows: And said matrix selection means, wherein each column generates an extended first processing block of L columns having N pixels whose pixel value is the value of the element of each said extended first vector. Processing block generating means for generating, for each row of the extended first processing block, each element generates an L-dimensional second vector that is a pixel value included in each row of the extended first processing block. Vector generating means; second selecting means for selecting one of the plurality of predetermined expansion matrices based on the M value and the N value; the expansion matrix selected by the second selecting means; N extended rows of said processing blocks, each row consisting of pixel values of N pixels, each element being a pixel, by generating N extended second vectors by multiplication with a second vector. From A video signal encoding device, comprising: