JPH10336653A - Image coder and image decoder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To optimize decoding processing by scalar multiplying with a value proportional to the square root of a pixel number for each image area with a conversion coefficient and giving the product to a quantization means to minimize total distortion of a decoded image. SOLUTION: An image signal is given to a block processing circuit 2, its area information is fed to a block processing circuit 1, where they are divided into block signals and block area information with respect to a plurality of blocks of a prescribed size of N×N. The block area information subject to block processing by the block processing circuit 1 is fed to support area correction circuits 4, 5, a coefficient introduction circuit 6a, and a discrimination circuit 7. (N×N) conversion coefficients introduced by a region supported(RS)- discrete cosine transform(DCT) section 10 are fed to a multiplier 12, where all the conversion coefficients are multiplied by a scalar value of M<1/2> /N. The conversion coefficients subject to scalar multiple of a value M<1/2> /N proportional to the square root of coding object pixel number M are fed to a quantizer 14, from which a quantized index signal is outputted.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding apparatus and a decoding apparatus, and more particularly to an image coding apparatus and a decoding apparatus for performing high-efficiency coding by processing an image signal such as a television signal in units of image components. The present invention relates to an image decoding device that decodes encoded data from the image encoding device via a medium.

[0002]

2. Description of the Related Art Conventionally, high-efficiency coding of an image signal is performed by concentrating signal power on a specific transform coefficient by orthogonal transform using a horizontal or vertical correlation of the image signal to reduce a required number of bits. There are many things to reduce. In the high-efficiency coding using the orthogonal transform, one image is divided into a plurality of unit blocks each having a fixed shape composed of a fixed number of pixels, and conversion is performed for each block. Vectors are represented as weighted linear sums, the weights are quantized, and the corresponding numbers are transmitted via media.

In the conventional image coding represented by MPEG standardization, one image is treated only as a continuous data sequence irrespective of the image content, and the image waveform is more faithfully realized with a smaller number of bits. The goal is to reproduce (patterns). In such conventional encoding, usually, DCT (Discrete Cosine Transform) is used.
And the like, which basically assumes continuity, the non-stationaryness inherent in the image signal, such as the foreground and background, leads to a decrease in coding efficiency and decoding image quality. Conventionally, signal processing and handling of images are
In the case of a single image, or in the case of a moving image, it is handled collectively for each continuous frame. Therefore, performing signal processing and handling in smaller units such as image component units has not been considered.

Therefore, in recent years, studies on encoding for processing and encoding after dividing an image signal into a plurality of regions corresponding to respective image components have been actively conducted. In this encoding, it is possible to select and apply an optimal processing method for each area while avoiding the above-described non-stationarity, thereby enabling higher-performance encoding. In addition, processing can be performed in units of image components, and a video description with better handleability can be realized.

[0005] In the conventional image signal waveform encoding,
The image has been divided into rectangular blocks of a certain size and processed as represented by DCT coding.

However, in the case of dividing into a plurality of regions and processing each region, since the region shape is not generally rectangular,
A conventional method of dividing an image into rectangular blocks of a certain size and processing the image cannot be used as it is. Therefore, based on the conventional block-based encoding method, some extended methods for processing a pixel group having an arbitrary shape have been proposed.

As one of such techniques, a DCT transform base is masked by an image area of an image to be coded and multiplied by a scalar to derive a new transform base vector for each image area. Region support DCT (RS-DCT: Region Sup
port DCT) has been proposed by the present applicants (Yoshiaki Kaga, "Study of Arbitrary Shape Coding Using Region-Supported DCT," IEICE Technical Report IE 95-109 (Dec. 1995),
Japanese Patent Application No. 7-325971). The transform coefficient derived in this way can be decoded by a well-known two-dimensional inverse DCT.

As a method similar to the RS-DCT, in order to use a rectangular two-dimensional DCT, pixels that do not belong to the encoding target region in the rectangle are extrapolated by averaging the levels of the pixels in the region. RF-DCT (Region
Filling DCT) has also been proposed (Junichi Kimura: "Study of image coding by region division in block" Proc. PC
SJ, '95, 3-7, pp. 39-40 (Oct. 1)
995), Norio Ito, Hiroyuki Katata, Hiroshi Kusao: "Study of DCT Realization Method for Arbitrary Shape Area", Proc. PCS
J, '95, 5-2, pp. 77-78 (Oct. 19)
95)).

In the RS-DCT and RF-DCT, when, for example, two image areas are mixed in a unit block, block DCT is performed for both image areas. Therefore, the transform coefficients are derived for each image area by the size of the block (ie, by the number of pixels of the block including the two image areas). Therefore, when the transform coefficients for both image areas are combined, two blocks of coefficients are derived for one block, and it is necessary to transmit these two blocks of transform coefficients. in this way,
In the RS-DCT and the RF-DCT, when two image areas are mixed in a block, it is necessary to transmit a transform coefficient twice as many as the number of pixels apparently. However, in normal high-efficiency coding using DCT, only the transform coefficients that are visually important and correspond to low-frequency components that occupy most of the signal power are transmitted, so that the transform coefficients to be transmitted increase in appearance. This does not lead to an effective increase in the amount of transmitted information.

In these methods, a transform coefficient is derived from a pixel to be coded included in a block. On the decoding side, a block-shaped image is apparently restored by two-dimensional inverse DCT, and the separately transmitted coding is performed. Only the pixels in the corresponding area are selected by the mask pattern indicating the target area and used for the decoded image.

Although the above method is simple, continuity in and around the low-frequency region is guaranteed to some extent, but discontinuity occurs in the middle and high-frequency regions, and the inserted pixel level and, consequently, the pixel description method in the region There is no guarantee of optimality.

[0012]

SUMMARY OF THE INVENTION RS-DCT, RF-
In both DCTs, the number of transform bases is the number of pixels in a block (hereinafter N) regardless of the number of pixels to be encoded (hereinafter M).
× N). Therefore, apparently (N × N)
Are obtained. However, the number of conversions is M (M ≦ N
× N) is performed on the value of the pixel to be coded, so that at most M of the obtained transform coefficients are actually significant (non-zero).

In high-efficiency coding using DCT, transform coefficients are quantized to reduce the amount of information, and at the same time, coding distortion due to a quantization error at the time of quantization occurs. However, quantization noise is 0 for a pixel whose transform coefficient value is 0. Therefore, in the entire block including M pixels to be encoded, the quantization noise is equivalent to M pixels. That is,
The quantization noise power for one pixel to be encoded is Pe
Then, M significant transform coefficients are obtained for M pixels, resulting in M × Pe coding distortion for the block.

After a block-like decoded image of (N × N) pixels is once obtained by two-dimensional inverse DCT from the encoded data obtained by such DCT, only M of these pixels are obtained. Are selected and left using the mask pattern. Along with this processing, distortion due to quantization noise distributed to other than the selected pixel is also removed.

That is, distortion due to quantization noise left in a block is

[0016]

(Equation 1)

In the pixel unit,

[0018]

(Equation 2)

## EQU1 ##

From this equation, the amount of distortion for each pixel selected on the decoding side is proportional to the number M of pixels to be encoded in the block to which the pixel belongs.

In order to optimize the total amount of distortion of the decoded image under the condition that the amount of information to be transmitted is constant, that is, to realize the most efficient encoding of the transmission information and the distortion, it is necessary to use the pixel unit. It is necessary to make the amount of distortion uniform. Therefore, when the encoding is performed as described above, the distortion amount for each pixel differs for each block and each image depending on M, and is not optimized.

Therefore, the present invention has been made in view of the above points, and has as its object to provide an image encoding apparatus and an image decoding apparatus which solve the above-mentioned problems.

[0023]

In order to achieve the above object, in the apparatus according to the first aspect of the present invention, an image composed of a plurality of constituent elements is formed by a plurality of pixels of a predetermined size formed of a predetermined number of pixels. A pixel signal vector corresponding to at least one image region composed of a single component included in the block is divided into blocks, and an orthogonal transform base derived for each image region based on boundary information of the component. In an image encoding apparatus including a conversion unit that converts and outputs a vector as a weighted linear sum and a quantization unit that quantizes a signal from the conversion unit, the block includes a weighted conversion coefficient from the conversion unit. Optimizing means for optimizing the encoding process by scalar multiplying by a value proportional to the square root of the number of pixels for each image area and inputting the result to the quantizing means is provided. Characterized in that was.

Here, in the apparatus according to the second aspect of the present invention, an image composed of a plurality of constituent elements is divided into a plurality of blocks of a predetermined size composed of a predetermined number of pixels, and a single block included in the block is divided. A conversion unit for converting and outputting a pixel signal vector corresponding to at least one image region composed of one component as a weighted linear sum of orthogonal transformation base vectors derived for each image region based on boundary information of the component; And a quantization means for quantizing a signal from the conversion means, wherein the signal level of the pixel signal vector is proportional to the square root of the number of pixels for each of the image regions included in the block. Optimizing means for optimizing the encoding process by scalar multiplying the obtained value by the scalar multiplication and inputting the input to the conversion means.

Here, in the apparatus according to the present invention, an image composed of a plurality of constituent elements is divided into a plurality of blocks of a predetermined size composed of a predetermined number of pixels, and a single block included in the block is divided. A conversion unit for converting and outputting a pixel signal vector corresponding to at least one image region composed of one component as a weighted linear sum of orthogonal transformation base vectors derived for each image region based on boundary information of the component; And a quantizing means for quantizing a signal from the transforming means, wherein the quantizing means has a value inversely proportional to the square root of the number of pixels for each of the image regions included in the block. And a quantization means for optimizing the encoding process by multiplying the quantization step width by a scalar.

Here, in the apparatus according to the present invention, the orthogonal transform base vector masks a predetermined component of the discrete cosine transform base vector for each of the image regions based on the boundary information of the constituent elements. And a scalar multiplication to obtain a newly derived orthogonal transformation base vector for each image area.

According to a fifth aspect of the present invention, there is provided an apparatus according to the present invention, comprising a predetermined number of pixels which are output from the image encoding apparatus according to the first aspect and are divided by a predetermined size. Inverse quantization means for inversely quantizing encoded data obtained by encoding a pixel signal vector corresponding to at least one image region composed of a single component included in a block to obtain a weighting coefficient; An inverse decoding unit for performing weighted linear addition of orthogonal transformation base vectors derived for each image region based on boundary information and decoding a pixel signal vector corresponding to the image region, The weighting coefficient obtained by the inverse quantization is scalar-multiplied by a value inversely proportional to the square root of the number of pixels for each of the image regions included in the block, and the orthogonal transform basis vector is calculated. Characterized by comprising an optimization means for optimizing a decoding process by the weighted linear addition of.

Here, in the apparatus of the present invention described in claim 6, the image output from the image coding apparatus described in claim 2 is:
A coded data obtained by coding a pixel signal vector corresponding to at least one image region consisting of a single component included in a block divided by a predetermined size and configured by a predetermined number of pixels is inversely quantized and weighted by a weighting coefficient. Inverse quantization means for calculating the weighted linear addition of orthogonal transformation basis vectors derived for each of the image regions based on the boundary information of the components, and decoding pixel signal vectors corresponding to the image regions. And an image decoding apparatus comprising: means for scalar multiplying the output of the inverse transform means by a value inversely proportional to the square root of the number of pixels for each of the image areas included in the block to correspond to the image area. Optimizing means for optimizing a decoding process by decoding a pixel signal vector is provided.

Here, in the apparatus of the present invention described in claim 7, the image output from the image coding apparatus according to claim 3 is:
A coded data obtained by coding a pixel signal vector corresponding to at least one image region consisting of a single component included in a block divided by a predetermined size and configured by a predetermined number of pixels is inversely quantized and weighted by a weighting coefficient. Inverse quantization means for calculating the weighted linear addition of orthogonal transformation basis vectors derived for each of the image regions based on the boundary information of the components, and decoding pixel signal vectors corresponding to the image regions. Means, wherein said inverse quantization means performs scalar multiplication of a quantization step width by a value proportional to a square root of the number of pixels for each of said image regions included in said block. It is characterized by comprising an inverse quantization means for optimizing the processing.

Here, in the apparatus according to the present invention, the orthogonal transform basis vector masks a predetermined component of the discrete cosine transform basis vector for each of the image areas based on the boundary information of the component. And a scalar multiplication to obtain a newly derived orthogonal transformation base vector for each image area.

[0031]

Embodiments of the present invention will be described below in detail with reference to the drawings.

(First Embodiment) FIG. 1 is a block diagram showing the configuration of a first embodiment of an image coding apparatus to which the present invention is applied.

In the present embodiment, the number of image components of the image to be processed is 2, and the image area in each divided unit block (hereinafter, referred to as a block) is 2 at the maximum (the minimum is 1). . Also, it is assumed that area information (boundary information) indicating an area occupied by each image component in the screen is known.

The image encoding apparatus shown in FIG. 1 includes an area support DCT unit 10, a multiplier 12, a quantizer 14, and a variable length encoder 16.

The area support DCT unit 10 includes a blocking circuit 1 for receiving area information, a blocking circuit 2 for receiving an image signal to be processed, and a DCT (Discrete Cosine T).
ransform; discrete cosine transform) basis supply circuit 3, support area correction circuits 4 and 5, coefficient derivation circuits 6a and 6
b and 6c, a combination circuit 9, a judgment circuit 7, and a switching circuit 8. The support area correction circuit is required for the number of image components of one processing target image.

The image signal is supplied to a blocking circuit 2 and its area information is supplied to a blocking circuit 1, where it is divided into block signals and block area information for a plurality of blocks of a fixed size of N × N. The value of the number of pixels N is generally 8 or 16.

Assuming that the input image is composed of, for example, image components of a person and a background, the block area information output from the blocking circuit 1 indicates that each block includes only a person or only the background as an image component, or And a part of the background as image components. In other words,
The block area information indicates that the image has a rectangular shape without a boundary of the image component in the block, or that the image has a non-rectangular shape with the boundary of the image component in the block.

The block area information thus blocked by the blocking circuit 1 is supplied to the support area correction circuits 4 and 5, the coefficient derivation circuit 6a, and the determination circuit 7. In the support area correction circuits 4 and 5, D
The support area of the DCT base supplied from the CT base supply circuit 3 is corrected for each block according to the image area shape.
The support region correction circuit 4 corresponds to the region shape of one image region (for example, a person) of the two image regions, and the support region correction circuit 5 corresponds to the region shape of the other image region (for the background). Modifications are made. The support area correction method by the support area correction circuits 4 and 5 will be described later.

The image signals blocked by the blocking circuit 2 are supplied to coefficient deriving circuits 6a, 6b and 6c. For the image signal of the block composed only of the person and the block composed only of the background, the two-dimensional DCT base supplied from the DCT base supply circuit 3 by the coefficient derivation circuit 6a is used without correction,
Derivation of the T coefficient and its encoding are performed. Coefficient derivation circuit 6
The encoded data from a is supplied to the switching circuit 8.

For an image signal of a block in which a part of each of the two image components is mixed, a coefficient deriving circuit 6
DCT using the modified DCT bases supplied from support region modifying circuits 4 and 5 by b and 6c, respectively.
Derivation of coefficients and encoding thereof. The derivation of this DCT coefficient and its encoding will be described later.

The encoded data from the coefficient deriving circuits 6b and 6c are supplied to a combination circuit 9, respectively. Then, encoded data describing two image regions constituting one block is multiplexed encoded data,
The signal is supplied from the combination circuit 9 to the switching circuit 8. The multiplexing of the coded data by the combination circuit 9 is a process for combining the coded data of two image areas in one block in order to perform switching on a block basis by the switching circuit 8, and the coefficient deriving circuits 6b and 6c 2 from
One coded data sequence is simply arranged in series.

On the other hand, in the judgment circuit 7, the blocking circuit 1
It is determined whether there is only one image region (person or background) or two image regions (person and background) are mixed for each block based on the block region information of each block supplied from. The determination signal from the determination circuit 7 is
It is supplied to the switching circuit 8.

The switching circuit 8 converts the encoded data from the coefficient deriving circuit 6a into a block in which two image areas coexist based on the determination signal from the determination circuit 7 for a block in which only one image area exists. On the other hand, the coded data from the combination circuit 9 is selectively switched and output as compressed serial data. That is, serially multiplexed encoded data from the combination circuit 9 is output to a block in which two image areas are mixed, and the coefficient derivation circuit 6a is output to a block composed of another one image area. And outputs encoded data represented by a weighted linear sum.

Here, the support area correction circuits 4 and 5
The method of correcting the support region according to the above and the method of deriving the coefficients in the coefficient deriving circuits 6b and 6c will be briefly described.

In the support area correction, a predetermined size N × N supplied from the DCT base supply circuit 3 is used.

[0046]

[Outside 1]

Is supplied.

N × N of the DCT orthogonal transform base vector
Of each of the components, M components corresponding to the encoding target (support) image area are held, and the other components are set to 0.

[0049]

[Outside 2]

Is set, and correction is made in the image area to be encoded. Then, the following equation (1) is calculated.

[0051]

[Outside 3]

Then, the masked vector is divided by the square of the norm to calculate a new transformed base vector corrected to the shape of the image component to be encoded, and the norm is corrected. That is,

[0053]

(Equation 3)

By

[0055]

[Outside 4]

Is required. The new transformed basis vector corrected to the shape of the image component after the norm correction is a scalar multiplication of each component of the masked vector corresponding to the encoding target pixel.

Next, a coefficient deriving method in the coefficient deriving circuits 7 and 8 will be described.

Here,

[0059]

[Outside 5]

Is an image signal vector (the number of elements N × N) from the blocking circuit 2.

First, in the first stage, the image signal vector and the corrected

[0062]

[Outside 6]

The scalar function is calculated by calculating the inner product of

[0064]

(Equation 4)

Converted to modified

[0066]

[Outside 7]

Then, the coefficient c _k having the maximum absolute value is selected from these coefficients by the following equation.

[0068]

(Equation 5)

Next, in the second stage, the coefficient corresponding to the coefficient c _k having the maximum absolute value is determined.

[0070]

[Outside 8]

[0071] The components of the product of the DCT orthogonal transform basis vector absolute value maximum coefficient c _k and the absolute value of the following equation corresponding to the maximum of the coefficient c _k by removing subtracted from the image signal vector A residual signal vector is obtained, and this is replaced with an image signal vector.

[0072]

(Equation 6)

Next, in the third step, the inner product of the image signal vector replaced with the above-mentioned residual signal vector obtained in the second step and the new transformed base vector corrected after the above-mentioned norm correction is obtained. Is calculated again in the same manner as in the first step, is converted into a scalar function, and a new coefficient having a maximum absolute value is selected from the coefficients of this function as in the first step.

Next, in the fourth stage, the component of the product of the coefficient c _k having the largest absolute value selected in the third stage and the DCT orthogonal transform base vector corresponding to this coefficient is subtracted from the image signal vector. Then, a new residual signal vector is obtained again in the same manner as in the second stage, and is replaced with an image signal vector.

Next, in the fifth stage, the residual signal power based on the residual signal vector is previously set to a sufficiently small value based on the residual signal vector obtained in the fourth stage, that is, the image signal vector. It evaluates whether it is equal to or less than the threshold value, and if it is equal to or less than the threshold value, adopts the corrected coefficient of the new transformed base vector at this time, encodes it as a weighted linear sum, and outputs the result. Here, the evaluation of the residual signal power is performed only for the image component to be coded. That is, the second
Is performed on N × N elements, but the purpose is to represent the level of M pixels to be coded. Therefore, the evaluation of the error is based on the elements of the residual signal vector. Of these, the sum of the squares of the elements corresponding to the M elements is used.

On the other hand, when it is determined that the residual signal power is not below the sufficiently small threshold value, the process returns to the third stage to continue the processing, and in the fifth stage, the residual signal power is previously sufficiently small. The processes of the third stage, the fourth stage, and the fifth stage are repeatedly executed until it is evaluated that the difference is equal to or smaller than the threshold value determined as the value.

Note that since the transformed base vectors are not orthogonal, the component once removed may reappear with the removal of other components. In this case, the coefficients of the same transformation basis are accumulated, and the result is finally adopted.

As described above, according to the RS-DCT unit 10, it is possible to handle an image signal in units of image components with high efficiency on a relatively small hardware scale.

(N) derived by the RS-DCT unit 10
× N) transform coefficients are supplied to the multiplier 12,

[0080]

[Outside 9]

The quantization index signal is supplied to the quantizer 14 and is output. The output index signal is supplied to the variable length encoder 16 and converted into a binary encoded data string, and transmitted via a transmission medium such as a communication medium or a recording medium.

With this encoding, the quantization noise power of one pixel to be encoded is represented by Pe regardless of the scalar-multiplied coefficient before quantization, and the coding distortion of the entire block due to the quantization noise power is obtained. Is represented by M × Pe.

The encoded data sequence from the transmission medium is decoded by the image decoding device shown in FIG.

In this decoding process, the coded data sequence is supplied to the variable length decoder 20, and the decoded quantization index is output. The output decoding quantization index is supplied to the inverse quantizer 22, and the inverse quantization transform coefficient is output. This inverse quantization transform coefficient is

[0085]

[Outside 10]

That is, scalar multiplication is performed with a value inversely proportional to the square root of the number M of encoded pixels.

Here, each image component (object)
Is known, and the number M of pixels to be encoded in each block is a DCT.
When deriving the coefficients, those already obtained are used.

The inversely quantized transform coefficient multiplied by scalar is N ×
The signal is supplied to the N two-dimensional inverse DCT unit 26, and (N × N) pixel signal levels are restored. in this way,

[0089]

[Outside 11]

The distortion due to the quantization noise power in the entire block of the restored pixel signal is as follows.

[0091]

(Equation 7)

Is obtained.

The restored (N × N) pixel signals are supplied to the selector 28. The selector 28 selects M pixels corresponding to the image area of the block from all the pixels in the (N × N) blocks using the mask pattern 29 indicating the image area in the block separately transmitted. And output. As a result, the final distortion amount for the entire block is

[0094]

(Equation 8)

Is obtained.

Therefore, the amount of coding distortion due to quantization noise in pixel units is inversely proportional to the number M of pixels to be coded.
The final distortion amount in pixel units cancels out the proportionality with the number M of encoding target pixels in RS-DCT, and becomes Pe independently of M, thereby uniformly optimizing the distortion amount in pixel units. can do.

As described above, the multiplier 12 is provided before the quantizer in the image encoding apparatus.

[0098]

[Outside 12]

The information amount allocation is optimized so that the amount of distortion in the decoded pixel unit becomes uniform, and the total amount of distortion of the decoded image when the amount of information to be transmitted is constant can be minimized. There is an effect that efficient transmission of encoded data can be realized.

(Other Embodiments) The processing by the multipliers 12 and 24 in the above embodiment is a scalar multiplication.
It is invariant to linear transformation. Therefore, even when the scalar multiplication by the multiplier is performed on the pixel signal level, it is equivalent to the scalar multiplication performed on the transform coefficient in the above embodiment.

Therefore, a multiplier is provided at the input or output of the blocking circuit 2 in the encoding device, and the pixel signal level is multiplied by the same scalar as that of the multiplier 12 in FIG. A multiplier is provided at the output of the two-dimensional inverse DCT unit 26, and the same effect as in the first embodiment can be obtained by performing a scalar multiplication on the pixel signal level in the same manner as the multiplier 24 in FIG. be able to.

Also, without providing a multiplier, the amount of distortion for each pixel is optimized uniformly by utilizing the fact that the quantization noise power increases in proportion to the square of the quantization step width.
It is also possible to configure so as to obtain the same effect as that of the embodiment.

That is, in the encoding device, the quantizer 1
4 is corrected by scalar multiplication by a value inversely proportional to the square root of the number M of pixels to be encoded in the RS-DCT unit 10, and further, the quantization step width of the inverse quantizer 22 in the decoding device. An equivalent effect can also be obtained by correcting scalar by a value proportional to the square root of the number M of pixels to be encoded.

However, since the quantization process is a division opposite to the multiplication, the quantization step width in the encoding device is the same as that in the first embodiment.

[0105]

[Outside 13]

It is necessary to perform scalar multiplication by using

[0107]

As described above, according to the present invention, it is possible to decode encoded data output from an encoding device that has optimized encoding based on boundary information of an image composed of a plurality of components. When decoding is performed by the device, an effect is obtained that the decoding process can be optimized so as to minimize the total distortion amount of the decoded image.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a first embodiment of an image encoding device to which the present invention has been applied.

FIG. 2 is a block diagram illustrating a configuration of a first embodiment of an image decoding device to which the present invention has been applied.

[Explanation of symbols]

 1, 2 Blocking circuit 3 DCT base supply circuit 4, 5 Support area correction circuit 6a, 6b, 6c Coefficient derivation circuit 7 Judgment circuit 8 Switching circuit 9 Combination circuit 10 RS-DCT unit 12, 24 Multiplier 14 Quantizer 16 Variable length encoder 20 Variable length decoder 22 Dequantizer 26 Two-dimensional inverse DCT unit 28 Selector 29 Mask pattern

 ──────────────────────────────────────────────────続 き Continued on front page (72) Inventor Yutaka Tanaka 1-10-11 Kinuta, Setagaya-ku, Tokyo Japan Broadcasting Corporation Broadcasting Research Institute

Claims (8)

[Claims] 1. An image composed of a plurality of constituent elements is divided into a plurality of blocks each having a predetermined size and composed of a predetermined number of pixels.
A pixel signal vector corresponding to at least one image region composed of a single component included in the block is obtained by weighting a linear sum of orthogonal transformation base vectors derived for each image region based on boundary information of the component. In an image encoding apparatus comprising: a conversion unit that converts and outputs as; and a quantization unit that quantizes a signal from the conversion unit, a weighted conversion coefficient from the conversion unit is provided for each of the image regions included in the block. An image encoding apparatus, comprising: an optimizing means for optimizing an encoding process by multiplying a scalar by a value proportional to the square root of the number of pixels and inputting the result to the quantizing means. 2. An image composed of a plurality of constituent elements is divided into a plurality of blocks of a predetermined size composed of a predetermined number of pixels,
A pixel signal vector corresponding to at least one image region composed of a single component included in the block is obtained by weighting a linear sum of orthogonal transformation base vectors derived for each image region based on boundary information of the component. In an image encoding apparatus comprising: a conversion unit that converts and outputs a signal from the conversion unit; and a quantization unit that quantizes a signal from the conversion unit, the signal level of the pixel signal vector is set for each of the image regions included in the block. An image encoding apparatus, comprising: an optimizing means for optimizing an encoding process by multiplying a scalar by a value proportional to the square root of the number of pixels and inputting the result to the conversion means. 3. An image composed of a plurality of constituent elements is divided into a plurality of blocks of a predetermined size composed of a predetermined number of pixels,
A pixel signal vector corresponding to at least one image region composed of a single component included in the block is obtained by weighting a linear sum of orthogonal transformation base vectors derived for each image region based on boundary information of the component. In an image encoding apparatus comprising: a conversion unit configured to convert and output as; and a quantization unit configured to quantize a signal from the conversion unit, wherein the quantization unit determines the number of pixels for each of the image regions included in the block. An image coding apparatus characterized in that the image coding apparatus is a quantization means for optimizing a coding process by scalar multiplying a quantization step width by a value inversely proportional to a square root. 4. The orthogonal transformation basis vector masks a predetermined component of a discrete cosine transformation basis vector for each image region based on boundary information of the component, and performs scalar multiplication to newly generate a component for each image region. 4. The derived orthogonal transformation basis vector.
The image encoding device according to any one of the above. 5. An image output from the image encoding apparatus according to claim 1, wherein at least one image area composed of a single component included in a block composed of a predetermined number of pixels and divided by a predetermined size. Inverse quantization means for inversely quantizing the encoded data obtained by encoding the corresponding pixel signal vector to obtain a weighting coefficient, and an orthogonal transform base vector derived for each image region based on the boundary information of the components. An inverse decoding means for performing a weighted linear addition and decoding a pixel signal vector corresponding to the image area, wherein the weighting coefficient obtained by the inverse quantization is the image included in the block. For each region, the decoding process is optimized by scalar multiplication by a value inversely proportional to the square root of the number of pixels and weighted linear addition of the orthogonal transform base vectors. Image decoding apparatus comprising the means. 6. An image encoding apparatus according to claim 2, wherein at least one image area composed of a single component included in a block composed of a predetermined number of pixels and divided by a predetermined size is output. Inverse quantization means for inversely quantizing the encoded data obtained by encoding the corresponding pixel signal vector to obtain a weighting coefficient, and an orthogonal transform base vector derived for each image region based on the boundary information of the components. A weighted linear addition, and an inverse transform unit for decoding a pixel signal vector corresponding to the image region, wherein the output of the inverse transform unit is calculated for each image region included in the block. Optimizing means for optimizing a decoding process by scalar multiplying by a value inversely proportional to the square root of the number of pixels and decoding a pixel signal vector corresponding to the image area; Image decoding apparatus characterized by. 7. An image output from the image encoding apparatus according to claim 3, wherein at least one image area including a single component included in a block composed of a predetermined number of pixels and divided by a predetermined size is included. Inverse quantization means for inversely quantizing the encoded data obtained by encoding the corresponding pixel signal vector to obtain a weighting coefficient, and an orthogonal transform base vector derived for each image region based on the boundary information of the components. A weighted linear addition, and an inverse transform unit for decoding a pixel signal vector corresponding to the image region, wherein the inverse quantization unit is configured for each of the image regions included in the block. An image decoding device, characterized in that it is an inverse quantization means for scalar multiplying a quantization step width by a value proportional to the square root of the number of pixels to optimize decoding processing. 8. The orthogonal transform basis vector masks a predetermined component of a discrete cosine transform basis vector for each of the image regions based on boundary information of the components, and performs scalar multiplication to newly generate a discrete component for each of the image regions. 8. The derived orthogonal transformation basis vector, wherein:
The image decoding device according to any one of the above.