JPH06113326A - Picture coder and picture decoder - Google Patents

Abstract

PURPOSE:To decode a highly compressed picture by using a prediction method applied to an optional component among plural components of picture elements of the picture to allow a prediction means to predict other component of the picture elements at least. CONSTITUTION:The device is provided with a prediction section 3 as a prediction means predicting a picture and generating a predicted picture, an arithmetic operation section 4 as a coding means coding the picture by using the prediction picture generated by the prediction section 3 and a VLC circuit 9. Then the prediction section 3 uses the prediction method applied to one optional component among plural components of picture elements of a picture to predict at least other components of the picture elements. For example, the prediction means uses the prediction method applied to an optional component such as Y component to predict the other U, V components among the three components Y, U, V.

Description

Detailed Description of the Invention

[0001]

FIELD OF THE INVENTION The present invention compresses and encodes images,
Further, the present invention relates to an image encoding device and an image decoding device suitable for use in decompression decoding.

[0002]

2. Description of the Related Art In a conventional image coding apparatus for compressing and coding an image, for example, a plurality of predictors having different prediction methods for predicting an image to be coded and generating a predicted image are provided, and each predictor is used. The difference between the predicted image predicted by the device and the image to be coded is calculated, that is, the prediction error is calculated, and the one with the smallest prediction error is coded.

[0003]

By the way, in order to be able to decode the data coded as described above, such an image coding apparatus has a prediction error with the image to be coded. Prediction information about the predictor that has generated the predicted image to be minimized is added as an overhead of encoded data.

Therefore, each pixel of the image is, for example, Y, U,
And a plurality of components such as V, a prediction value for each of the three components of Y, U, or V of each pixel is generated by the predictor, and three components of Y, U, or V of each pixel are generated. Since the prediction information in each is added to the data encoded as the overhead, there is a problem that the code amount increases and the compression efficiency deteriorates.

Therefore, there is a method in which the predictor in the image coding apparatus is only one that predicts a predicted image only by a predetermined prediction method. However, with this method,
Depending on the image, the prediction error becomes large and the compression efficiency also deteriorates.

Further, in order to reduce the prediction information as an overhead, the predictor is not switched for each component forming a pixel, but the predictor is switched for each predetermined block unit, that is, a block is configured. There is a way to predict all pixels (Y, U, and V components) with the same predictor. However, this method has a problem that the prediction error becomes large and the compression efficiency is deteriorated as compared with the case where the predictor is switched for each component that constitutes a pixel.

The present invention has been made in view of such a situation, and improves the compression efficiency of an image.

[0008]

An image coding apparatus according to claim 1 predicts an image, and a prediction unit 3 as a prediction unit for generating a prediction image, and a prediction image generated by the prediction unit 3. An arithmetic unit 4 and a VLC circuit 9 as an encoding unit for encoding an image are provided, and the prediction unit 3 is applied to any one of a plurality of components forming pixels of an image. The above prediction method is characterized in that at least another component of a pixel is predicted.

An image coding apparatus according to a second aspect is characterized in that the pixels of the image are composed of three components of Y, U, and V.

An image coding apparatus according to a third aspect is characterized in that a pixel of an image is composed of three components of R, G and B.

An image coding apparatus according to a fourth aspect is characterized in that a pixel of an image is composed of four components of Y, M, C and K.

An image coding apparatus according to a fifth aspect is characterized in that the pixels of the image are composed of three components of Y, I, and Q.

An image decoding apparatus according to a sixth aspect of the present invention decodes an image using a prediction unit 25 as a prediction unit that predicts an image and generates a predicted image, and the predicted image generated by the prediction unit 25. A prediction method that includes an inverse VLC circuit 22 and a calculator 28 as a decoding unit for converting into a prediction unit, and applies the prediction unit 25 to any one of a plurality of components that form pixels of an image. At least another component of the pixel is predicted.

[0014]

In the image coding apparatus having the above structure, the prediction section 3 predicts an image and a predicted image is generated. Then, the image is encoded using the predicted image generated by the prediction unit 3. At this time, for example, three components of Y, U, and V (three components of R, G, and B) that configure the pixels of the image are included in the prediction unit 3.
(Four components Y, M, C, and K)
A prediction method applied to any one of (three components of Y, I, and Q), for example, a Y component is used to predict at least U and V components as other components of a pixel. . Therefore, the compression efficiency of the image can be improved.

In the image decoding apparatus of the present invention, the prediction unit 25 predicts an image and generates a predicted image.
Then, the image is decoded using the predicted image generated by the prediction unit 25. At this time, the prediction unit 25 is caused to predict at least another component of the pixel by the prediction method applied to any one of the plurality of components forming the pixel of the image. Therefore, a highly compressed image can be decoded.

[0016]

1 is a block diagram showing the configuration of an embodiment of an image coding apparatus according to the present invention. Delay circuit 1 or 2a
2d to 2d represent, for example, Y, U, or V components that constitute pixels of an image, at predetermined times T ₁ , T _2a , T
_2b , T _2c , or T _2d, respectively.

Here, the pixels A, B, and B of the array shown in FIG.
If C, D, or X is sequentially input to the delay circuits 1 and 2a to 2d, when the pixel X input later in time is output after being delayed by the delay circuit 1, the pixel X is output in time. The delay circuit 1 is provided so that the pixels A to D which have been previously input and have already been delayed by the delay circuits 2a to 2d are output.
Alternatively, the delay time T ₁ , T _2a , T _2b , T _2c , or T _{2d in} each of the delay circuits 2a to _2d is adjusted in advance.

Further, the Y, U, or V component constituting the pixel is expressed by the formula (1) from the R (red), G (green), and B (blue) components as the three primary colors of light constituting the pixel. ).

[0019]

[Equation 1]

The predictors 3a to 3d of the predictor 3 are, for example, as shown in FIG. 3, multipliers 11a to 11d for multiplying the input signal by a to d, and a multiplier 11a.
From the pixels A to D (Y, U, or V components) output from the delay circuits 2a to 2d to the pixel X (Y from the delay circuit 1). , U, or the predicted value X _'a, X' of the V component), _b, to generate X _'c, or X' _d, respectively.

The calculator 4a to 4d of the arithmetic unit 4, the prediction value X _'a, X' of the pixel X which is generated by the predictor 3a to 3d _b,
The prediction error for the pixel X is calculated by taking the difference between the pixel X and X ′ _c or X ′ _d .

The minimum value detection circuit 5 is composed of arithmetic units 4a to 4d.
When the prediction error calculated by the above is for the Y component of the pixel X, the minimum prediction error of the absolute value among them is detected and output to the terminal 8a of the switch 8. At the same time, the minimum value detection circuit 5 sets the prediction value set in advance in the predictor (one of the predictors 3a to 3d) in which the prediction value used when calculating the minimum prediction error of the detected absolute value is generated. Unit number (in the figure, predictors 3a to 3)
The number given in parentheses to d) is detected and output to the register 7.

The register 7 temporarily stores the predictor number output from the minimum value detection circuit 5 and outputs it to the selector 6 and the multiplexing circuit 10.

The selector 6 uses the predicted value of the U or V component of the pixel X generated by the predictor (one of the predictors 3a to 3d) corresponding to the predictor number output from the register 7, The prediction error for each U or V component of the pixel X calculated by the arithmetic unit (any of the arithmetic units 3a to 3d) is selected and output to the terminal 8b of the switch 8.

The switch 8 has a terminal 8 at a predetermined timing.
The prediction error for the Y component of the pixel X output from the minimum value detection circuit 5 and the prediction error for the U or V component of the pixel X output from the selector 6, which are alternately switched to the a side or the terminal 8b side. Is supplied to the VLC circuit 9.

The VLC circuit 9 supplies the prediction error for the Y component of the pixel X supplied from the minimum value detection circuit 5 via the switch 8 or the prediction error for the U or V component of the pixel X supplied by the selector 6, respectively. VLC (Variable Length Coding) processing is applied to the error, and V
The LC code is output to the multiplexing circuit 10.

The multiplexing circuit 10 multiplexes the VLC code output from the VLC circuit 9 with the predictor number supplied from the register 7 as its overhead, and outputs data in the format shown in FIG. 4, for example. .

Next, the operation will be described. The Y, U, or V components that make up each pixel of the image are
For example, the signals are sequentially input to the delay circuits 1 and 2a to 2d in this order, and are delayed there.

Now, as of now, from the delay circuit 1 to the time T ₁
If the pixel X (Fig. 2) delayed by only
The pixel A (Y, U, or V component) adjacent to the upper left of the pixel X, which is input to the delay circuits 2a to 2d before the time T _2a , T _2b , T _2c , or T _2d , above the pixel X A pixel B (Y, U, or V component) adjacent to the pixel X, a pixel C (Y, U, or V component) adjacent to the upper right of the pixel X, or a pixel D (Y, U, or V component) (FIG. 2) is output from each.

In the predictors 3a to 3d, the pixels A to D (Y, output from the delay circuits 2a to 2d are output.
U, or V component), a pixel output from the delay circuit 1 X (Y, U, or the predicted value X _'a, X' of the V component) _b, X _'c or X,' _d are respectively generated .

The predictors 3a to 3d are, for example, the pixel D (Y, U, or V component forming the pixel D), the pixel B (Y, U, or V forming the pixel B).
Component), pixel B (Y, U constituting pixel B,
Alternatively, the average value of the V component) and the pixel D (the Y, U, or V component forming the pixel D), or the pixel B (the Y, U, or V component forming the pixel B) and the pixel D (forming the pixel D). to Y, U or from the sum of the V component), Y constituting the pixel a (pixel a, U, or the minus the V component), the predicted value X _'a, X' of the pixel X _b, X ' It is designed to be output as _c or _X'd respectively.

[0032] _{Thus, X 'a = D X'} b = B X 'c = (B + D) / 2 X' d = B + because it is D-A, the multipliers 11a to 1 incorporating predictor 3a
1d (FIG. 3): a = 0, b = 0, c = 0, d = 1 Multipliers 11a to 11d built in the predictor 3b (FIG. 3)
, A = 0, b = 1, c = 0, d = 0 multipliers 11a to 11d built in the predictor 3c (FIG. 3)
, A = 0, b = 1/2, c = 0, d = 1/2 Multipliers 11a to 11d built in the predictor 3d (FIG. 3)
In, a = -1, b = 1, c = 0, d = 1 and a to d are set, respectively.

In the multipliers 11a to 11d (FIG. 3), a, b, c, or d can be set to any value, and the method of predicting the pixel X is limited to the above method. Not a thing.

The predicted value X _'a, X' of the pixel X which is generated by the predictor 3a to 3d _b, X _'c or X,' _d are outputted to the arithmetic unit 4a-4d. In computing unit 4a to 4d, the predicted value X _'a, X' of the pixel X output from the predictor 3a to 3d _b, X _'c or X,' and _d, the difference between the pixel X taken, pixel X The prediction error for each is calculated.

When the prediction error calculated by each of the arithmetic units 4a to 4d is for the Y component of the pixel X, the minimum value detection circuit 5 detects the minimum prediction error of the absolute values, and the switch 8 Is output to the terminal 8a. At the same time, in the minimum value detection circuit 5, the prediction value used when calculating the minimum prediction error of the detected absolute value is preset in the generated predictor (one of the predictors 3a to 3d). The predictor number (the numbers in parentheses attached to the predictors 3a to 3d in the figure) is detected and output to the register 7.

At the timing when the prediction error for the Y component of the pixel X is output from the minimum value detection circuit 5, the switch 8 switches to the terminal 8a side, and the prediction error for the Y component of the pixel X is input to the VLC circuit 9. It Then, in the VLC circuit 9, the prediction error for the Y component of the pixel X is subjected to VLC processing, and the VLC code is output to the multiplexing circuit 10.

On the other hand, in the register 7, the predictor number output from the minimum value detection circuit 5 is temporarily stored and output to the selector 6 and the multiplexing circuit 10.

In the selector 6, predictors (predictors 3a to 3a) corresponding to the predictor number output from the register 7 are output.
(any one of d), using the predicted value of the U or V component of the pixel X generated by
The prediction error for each of the U or V components of the pixel X calculated in any one of a to 3d) is selected and output to the terminal 8b of the switch 8.

That is, in the selector 6, prediction is performed by the same prediction method as the prediction method for predicting the prediction value of the Y component of the pixel X (the same predictor as the predictor in which the prediction value of the Y component of the pixel X is generated). Pixel), pixel X
The prediction error of the U or V component of the pixel X calculated using the prediction value of the U or V component of the pixel is selected.

At the timing when the selector 6 outputs the prediction error for the U or V component of the pixel X,
The switch 8 is switched to the terminal 8b side, and the prediction error for the U or V component of the pixel X is input to the VLC circuit 9 via the switch 8. And V
In the LC circuit 9, the prediction error for the U or V component of the pixel X is subjected to VLC processing, and each VLC code is output to the multiplexing circuit 10, respectively.

In the multiplexing circuit 10, the VLC code of the prediction error for the Y, U, or V component of the pixel X output from the VLC circuit 9 is multiplexed, and the predictor supplied from the register 7 The number is further multiplexed as its overhead, and the data of the format as shown in FIG. 4 is output.

As described above, when predictive coding of, for example, Y, U, or V components forming pixels of an image, of Y, U, or V components that are correlated with each other, for example, of the Y component, Since the prediction method of predicting the prediction value is the same as the prediction method of calculating the prediction value of the Y component (the same predictor as the one calculating the prediction value of the Y component), the prediction value of the U or V component as another component is predicted. , Y, U, or V component prediction values are predicted by different prediction methods (in this case, the same prediction method may occur by accident), that is, the prediction values of the Y, U, or V components are predicted. Compared to the case where information about a method is required, information about a method for predicting a predicted value as overhead (in the embodiment, a predictor number ) Is finished at 1/3, it is possible to improve the compression efficiency.

Further, in this case, since the human vision is not sensitive to the U or V component as the color component of the pixel, the U or V component can be thinned out in advance for coding, and therefore the compression efficiency can be improved. Can be further improved.

Next, FIG. 5 is a block diagram showing the configuration of an embodiment of an image decoding apparatus for decoding the image data encoded by the image encoding apparatus shown in FIG. The demultiplexing circuit 21 demultiplexes the VLC code of the prediction error for the pixel and the predictor number as the overhead, which are multiplexed as shown in FIG.
The LC code is output to the inverse VLC circuit 22 and the predictor number is output to the register 23.

The inverse VLC circuit 22 performs inverse VLC processing on the VLC code of the prediction error for the pixel output from the demultiplexing circuit 21, and outputs the prediction error for the pixel.

The delay circuit 24 delays the prediction error for the pixel output from the inverse VLC circuit 22 by the time T ₁ as in the delay circuit 1 of FIG. 1 and outputs it to the calculator 28.

The calculator 28 outputs the prediction error of the pixel output from the delay circuit 24 and the predictor 2 via the selector 26.
Any of the predicted values of the pixel output from 5a to 25d is added to output the decoded pixel.

The delay circuits 27a to 27d are provided in the arithmetic unit 28.
The output pixels are delayed by time T _2a , T _2b , T _2c , or T _2d, respectively, similarly to the delay circuits 3a to 3d in FIG. 1, and output to the predictors 25a to 25d.

Therefore, the pixels A, B, and B of the array shown in FIG.
Considering C, D, or X, in this image decoding apparatus, when the prediction error for the pixel X is output from the delay circuit 24, it is already decoded and the delay circuits 27a to 27 are used.
Pixel A, B, C or D latched (delayed) by d
But, it is designed to be output from each.

The predictors 25a to 25d of the predicting unit 25 are
The predictors 3a to 3d shown in FIG.
Pixels A to D output from the delay circuits 27a to 27d
(Y, U, or V component) to delay circuit 2
Prediction value X _'a, X' of the pixel X which is output from the 4 (Y, U or V component,) _b, to generate X _'c, or X' _d, respectively.

[0051] That is, the predictor 25a to 25d is the predicted value X _'a, X' of the pixel X to be generated by the predictor 3a to 3d in Figure 1 (Y, U or V component,) _b, X _'c, or X _'d identical to those of the generating respectively.

The register 23 temporarily stores the predictor number output from the demultiplexing circuit 21 and outputs it to the selector 26.

The selector 26 is a predictor (predictor 2) corresponding to the predictor number output from the register 23 (numbers in parentheses attached to the predictors 25a to 25d in the figure).
5a to 25d) pixel X (Y,
Predicted value of U or V component (predicted value _X'a ,
X _'b, X' either _c or X _'d,) and outputs it to the arithmetic unit 28.

In the image decoding apparatus configured as described above, for example, when the data in which the pixel X is encoded is input to the demultiplexing circuit 21, the data is demultiplexed there and the pixel X ( The VLC code of the prediction error for the Y, U, or V component) is output to the inverse VLC circuit 22, and the predictor number is output to the register 23.

In the inverse VLC circuit 22, the VLC code of the prediction error for the pixel X output from the demultiplexing circuit 21 is subjected to the inverse VLC processing, and the prediction error for the pixel X is output.

Then, in the delay circuit 24, the inverse VLC
The prediction error for the pixel output from the circuit 22 is shown in FIG.
Is the delay of the delay circuit 1 and only T ₁ time as well, the calculator 2
8 is output.

At the same time, in the delay circuits 27a to 27d, the X of the pixel X, which has been output from the arithmetic unit 28 and latched (delayed) by the arithmetic unit 28 before the time T _2a , T _2b , T _2c , or T _2d , has already been reached. Pixel A (Y, U, or V component) adjacent to the upper left of pixel, pixel B adjacent to above pixel X
(Y, U, or V component), pixel C (Y, U, or V component) adjacent to the upper right of pixel X,
Alternatively, the pixel D (Y, U, or V) adjacent to the left of the pixel X
Components) (FIG. 2) are each output from there.

The pixel A, B, C, or D output from each of the delay circuits 27a to 27d is the prediction circuit 25a.
To 25d, where the predictors 3a to 3 of FIG.
Similarly to d, the pixel X output from the delay circuit 24
(Y, U or V component) predicted value X _'a of,
_X'b , _X'c , or _X'd is generated, respectively.

That is, in the predictors 25a to 25d,
Pixel D (Y, U, or V component forming pixel D), pixel B (Y, U, or V component forming pixel B), pixel B (Y forming U, Y, or
U or V component) and the average value of pixel D (Y, U, or V component forming pixel D),
Alternatively, from the sum of the pixel B (Y, U, or V component forming the pixel B) and the pixel D (Y, U, or V component forming the pixel D), the pixel A (Y, U forming the pixel A, Alternatively, the value obtained by subtracting the V component is the pixel X (Y, U, or V that constitutes the pixel X).
Prediction value X _'a, X' component) _b, it is output as X _'c or X,' _d.

On the other hand, in the register 23, the predictor number output from the demultiplexing circuit 21 is temporarily stored and output to the selector 26.

In the selector 26, the predictor number output from the register 23 (the predictors 25a to 25d in the figure)
Pixel X generated by the predictor (one of the predictors 25a to 25d) corresponding to the number in parentheses)
A predicted value (any one of predicted values X ′ _a , X ′ _b , X ′ _c , or X ′ _d ) of (Y, U, or V component) is selected and output to the calculator 28.

In the calculator 28, the prediction error of the pixel X (Y, U, or V component of the pixel X) output from the delay circuit 24 and one of the predictors 25a to 25d selected by the selector 26 are selected. Pixel X output by
The predicted value of (Y, U, or V component of pixel X) is added, and the decoded pixel X (Y, U, or V component of pixel X) is output.

In the present embodiment, the number of predictors that generate (calculate) the predicted value of the pixel is four, but the number of predictors is not limited to this.

Furthermore, in the present embodiment, the predicted value of the pixel X is generated (calculated) from the four pixels A to D (FIG. 2) adjacent to the pixel X, but, for example, the pixels A, B, and In addition to D and the pixels A to D, the predicted value of the pixel X can be generated (calculated) from the pixel adjacent to the left of the pixel D.

Further, in this embodiment, Y, which is calculated from the R, G, and B (blue) components as the three primary colors of light forming the pixel according to the equation (1),
Although the U and V components are used, the present invention is not limited to this. For example, the R, G, and B components and the Y, I, and Q components calculated from them according to the equation (2) or (3), respectively. Alternatively, Y, M, C, and K components can be used.

[0066]

[Equation 2]

[0067]

[Equation 3]

Here, Y, M calculated from the equation (3),
Alternatively, if the density of the C component is 0.7, 0.5, or 0.9, as shown in FIG. 6A, for example, the common parts of the C, M, or Y components are so-called This is a K component (Fig. 6 (b)) because it becomes black by the basic principle of subtractive color mixing.
(And can be)

Further, in the present embodiment, when predictive coding of, for example, Y, U, or V components forming pixels of an image, among Y, U, or V components having a correlation with each other, for example, Predict the predicted value of the U or V component as another component by the same prediction method as the predicted value of the Y component (with the same predictor as the predictor that calculated the predicted value of the Y component) However, it is possible to use the same prediction method to predict not only the component that constitutes itself, but also the Y, U, or V component of a pixel that is relatively correlated with the pixel, such as a pixel adjacent to the pixel. it can.

[0070]

As described above, according to the image coding apparatus of the present invention, the image is predicted by the prediction means, and the predicted image is generated. Then, the image is encoded using the predicted image generated by the prediction means. At this time, for example, three components of Y, U, and V (three components of R, G, and B) (four components of Y, M, C, and K) that configure the pixels of the image are included in the prediction unit. ) (3 components Y, I, and Q)
The prediction method applied to any one of them, for example, the Y component, causes at least the U and V components as other components of the pixel to be predicted. Therefore, the compression efficiency of the image can be improved.

In the image decoding apparatus of the present invention, the image is predicted by the prediction means and the predicted image is generated. Then, the image is decoded using the predicted image generated by the prediction means. At this time, the prediction means is made to predict at least another component of the pixel by the prediction method applied to any one of the plurality of components forming the pixel of the image. Therefore, a highly compressed image can be decoded.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of an image encoding device of the present invention.

FIG. 2 is a plan view showing an arrangement of pixels.

FIG. 3 is a diagram showing more details of a prediction circuit 3a (3b, 3c, or 3d) of the embodiment of FIG.

4 is a diagram showing a format of data output from the multiplexing circuit 10 of FIG.

FIG. 5 is a block diagram showing a configuration of an embodiment of an image decoding apparatus of the present invention.

FIG. 6 is a diagram for explaining a method of calculating a K component from Y, M, and C components.

[Explanation of symbols]

 1, 2a to 2d Delay circuit 3 Prediction unit 3a to 3d Predictor 4 Operation unit 4a to 4d Operation unit 5 Minimum value detection circuit 6 Selector 7 Register 8 switch 9 VLC circuit 10 Multiplexing circuit 11a to 11d Multiplier 12 Operation unit 21 Demultiplexing circuit 22 Inverse VLC circuit 23 Register 24 Delay circuit 25 Prediction unit 25a to 25d Predictor 26 Selector 27a to 27d Delay circuit 28 Operation unit

Claims (6)

[Claims] 1. A predicting unit that predicts an image and generates a predictive image, and the predictive image generated by the predicting unit,
Coding means for coding the image, the prediction means is a prediction method applied to any one of a plurality of components constituting the pixel of the image, at least other than the pixel An image coding apparatus characterized by predicting a component of a. 2. The image coding apparatus according to claim 1, wherein the pixels of the image are composed of three components of Y, U, and V. 3. The image coding apparatus according to claim 1, wherein the pixels of the image are composed of three components of R, G, and B. 4. The image coding apparatus according to claim 1, wherein the pixels of the image are composed of four components of Y, M, C, and K. 5. The image coding apparatus according to claim 1, wherein the pixels of the image are composed of three components of Y, I, and Q. 6. A predicting unit that predicts an image and generates a predictive image, and the predictive image generated by the predicting unit,
And a decoding unit that decodes the image, wherein the prediction unit is a prediction method applied to any one of a plurality of components forming the pixel of the image, An image decoding apparatus characterized by predicting a component of a.