JPH02131097A - Picture coder - Google Patents

Abstract

PURPOSE:To attain high speed coding by reading a component signal no thinned and a thinned component signal from a storage section in the unit of blocks comprising plural picture elements, sharing the signal to plural orthogonal conversion section and transferring the conversion coefficient to a quantization section. CONSTITUTION:A CB component signal is subjected to thinning in the horizontal direction and a CR component signal is subject to thinning in the horizontal direction respectively by thinning sections 3, 5, Y, CB and CR component signals are subject to discrete cosine transform in the unit of blocks and 8X8 conversion coefficients are obtained by 3 DCT conversion sections 8-10. The Y, CB and CR component signals are read form storage sections 2, 4, 6 in the unit of blocks so as to avoid idle conversion processing in the 3 DCT conversion sections 8-10, the result is shared to the 3 DCT conversion sections 8-10, which transfers the conversion coefficient to a quantization section 11. Thus, plural orthogonal conversion sections are used effectively to apply high speed coding.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image encoding device for shortening the transmission time of image signals or reducing the storage capacity.

[Conventional technology]

Multivalued color image (for example, 1 pixel 24 bits, 1677
Data compression methods for (7216 colors) include information-storing encoding and non-information-storing encoding. Information-preserving encoding refers to encoding that does not include quantization in the encoding process, and although it is possible to reproduce an image that is exactly the same as the original image through encoding and decoding, it does not require a high compression rate. cannot be obtained. On the other hand, non-information preserving encoding refers to encoding that includes some kind of quantization processing during the encoding process, and the reproduced image contains quantization noise due to the encoding/decoding process, resulting in deterioration of image quality. , a high compression ratio can be obtained.

In the case of non-information preserving encoding, quantization distortion (S/
It is evaluated based on the relationship between the S/N ratio (N ratio) and the data compression rate (information amount), but one method to achieve a good relationship between S/N ratio and information amount is to quantize the transform coefficients after orthogonal transformation and make them variable. There is a long encoding method.

Furthermore, in non-information preserving encoding, the RGB signal output from the reading system is separated into component signals of a luminance signal and a color difference signal, taking advantage of the fact that human visual perception is dull to color difference signals. Either perform thinning processing to compress the amount of information, or compress the amount of information by using different quantization characteristics for the luminance signal and color difference signal, and using coarser quantization characteristics for the color difference signal. In some cases, this may be attempted. When performing thinning processing on color difference signals, only low frequency components may be extracted using a low-pass filter before performing thinning processing.

In addition, when performing orthogonal transformation on a multivalued color image, quantizing it, and encoding it, the array of the encoded data can be formatted as a field-sequential type, line-sequential type, etc., with each block consisting of a plurality of pixels undergoing orthogonal transformation as a unit. There is the next type. Furthermore, when there is a demand for quickly displaying or recording blocks as color images, the encoded data is often arranged in a dot-sequential manner.

In addition, in this type of encoding device that performs orthogonal transformation on a multivalued color image and then quantifies and encodes it, it takes a lot of time to process the orthogonal transformation. Therefore, when high-speed processing is required, A total of three orthogonal transform units, one each for a luminance component (hereinafter referred to as Y component) and two types of color difference components (hereinafter referred to as CB component and CR acid component), are prepared and operated in parallel, In many cases, efforts are being made to speed up the processing.

[Problem to be solved by the invention]

In the color image signal encoding device described above, 3
The processing components are fixedly assigned to the orthogonal transform units. Here, when thinning processing is performed on the chrominance component signal, the number of pixels of the luminance component signal and the number of pixels of the chrominance component signal are different, and the number of pixels of the chrominance component signal is smaller than the number of pixels of the luminance component signal. Become. Therefore, when each component signal is divided into blocks each consisting of multiple pixels in order to perform orthogonal transformation, the number of blocks for which orthogonal transformation is performed is that the number of blocks for the chrominance component signal is smaller than the number of blocks for the luminance component signal. . For example, if the original image has X pixels in the horizontal direction,
If the vertical direction is made up of X pixels and the number of pixels of the color difference component signal is thinned out to half, the number of pixels of the luminance component signal is xXy pixels, and the number of pixels of the color difference component signal is (xXy
)/2 pixels. Furthermore, if the block size of the orthogonal transformation is n pixels in the horizontal direction and n pixels in the vertical direction, the number of blocks of the luminance component signal to be orthogonally transformed is (xxy)/(
nXn), and the number of blocks of the color difference component signal is (xXy)/
(2X (nxn)).

Therefore, in the image signal encoding device described above, when performing thinning processing on color difference component signals and encoding each component signal, three orthogonal transform units are fixedly allocated to each component signal. , during the orthogonal transformation processing of the luminance component signal, there is a time when the orthogonal transformation section of the chrominance component signal does not operate, or even though the orthogonal transformation of the chrominance component signal has been completed for the entire image, the luminance The orthogonal transformation of the component signals ends up being incomplete. In this way, the entire orthogonal transform processing time is determined by the processing time of the orthogonal transform section that handles the luminance signal, and there is a problem in that the high-speed processing effect of parallel processing cannot be fully demonstrated.

SUMMARY OF THE INVENTION An object of the present invention is to solve the above-mentioned problems and provide an encoding device that can effectively use a plurality of orthogonal transform units and perform high-speed encoding.

[Means to solve the problem]

The encoding device of the present invention includes: a color conversion unit that converts an image signal into a plurality of component signals; a thinning unit that thins out at least one component signal among the plurality of color-converted component signals; a storage unit that stores the component signal that has been thinned out and the component signal that has been thinned out for a plurality of pixels; a plurality of orthogonal transform units that obtain a plurality of transform coefficients, a quantization unit that quantizes the transform coefficients and outputs a quantization index, an encoding unit that encodes the quantization index, and the plurality of orthogonal transforms. The unthinned component signals and the thinned out component signals are read out from the storage section in blocks each consisting of a plurality of pixels, so that there is no vacancy in the conversion processing operation in the plurality of orthogonal transformation sections. and a control section that transfers the transform coefficients from the orthogonal transform section to the quantization section.

[Effect]

The key point of the encoding device of the present invention is to use orthogonal transformation,
Another object of the present invention is to effectively use a plurality of prepared orthogonal transform units in encoding in which a color difference signal is thinned out, and to encode an image signal at high speed.

Therefore, in the present invention, unthinned component signals and thinned out component signals are read out from the storage unit in blocks each consisting of a plurality of pixels, so that there is no vacancy in the transformation processing operations in the plurality of orthogonal transformation units. , and are distributed to a plurality of orthogonal transform units, and the transform coefficients are transferred from the orthogonal transform unit to the quantization unit.

〔Example〕

Hereinafter, the present invention will be explained in detail with reference to the drawings.

In the embodiment of the present invention, an R component signal and a G component signal.

A multivalued color image signal consisting of three components of the B component signal is
The signal is converted into a component signal, a CB component signal, and a CR component signal.

Next, pixels are thinned out from each of the image signals of the CB component signal and the CR component signal. Prior to this pixel thinning, the signal band may be limited to prevent aliasing distortion. As a method for thinning out pixels, pixels are often thinned out by half in the horizontal direction, or thinned out by half in both the horizontal and vertical directions. As a result of this pixel thinning, the number of pixels of the CB component signal and CR component signal is smaller than the number of pixels of the Y component signal.

Next, each component signal is read out in units of blocks each consisting of a plurality of pixels. As this block, a square block consisting of n×n pixels is often used. The number of blocks to be read for one image is determined by pixel thinning for the CB component signal and CR component signal, so the number of blocks to be read is determined by the Y component signal and
The CB component signal and the C acid component signal are different. For example, C.B.
Half of the component signal and CR component signal in the horizontal direction
If the number of blocks is thinned out, the number of blocks to be orthogonally transformed will be two blocks of the Y signal and one block each of the CB component signal and the CR component signal.

When three orthogonal transform units are prepared, blocks are read out as shown in FIG. 2 in order to ensure that each orthogonal transform unit has no processing time. In FIG. 2, 21 is the first block of the Y component signal, 22 is the second block of the Y component signal, 23 is the first block of the thinned out CB component signal, and 24 is the first block of the thinned out CR component signal. 1 block, 25
26 is the third block of the Y component signal, and 26 is the fourth block of the Y component signal.
Block 27 is the second block of the thinned out CB component signal, 28 is the second block of the thinned out CR component signal, 2
9 is the fifth block of the Y component signal, 30 is the sixth block of the Y component signal, 31 is the third block of the thinned out CB component signal, 32 is the third block of the thinned out CR component signal,
Reference numeral 33 indicates a block to be read out for the first time, 34 a block to be read out for the second time, 35 a block to be read out for the third time, and 36 a block to be read out for the fourth time.

As shown in FIG. 2, in the first reading, two blocks of Y component signals and one block of CBB signals are read out,
In the second reading, one block of CR component signal and one block of Y
Two blocks of component signals are read out, and in the third readout, one block of CBB component signals, one block of CR component signals, and one block of Y component signals are read out, and in the fourth readout, one block of Y component signals and one block of CBB component signals are read out. 1 block and 1 block of CR component signals are read out. Thereafter, the read processing is repeated in the same order. With one readout,
The three read blocks are distributed to three orthogonal transform units. In this way, when a plurality of orthogonal transform units are prepared, blocks corresponding to the number of prepared orthogonal transform units are read out at one time and transferred to each orthogonal transform unit at one time.

Then, orthogonal transformation is performed on each block to obtain a plurality of transform coefficients. As this orthogonal transformation, any orthogonal transformation such as two-dimensional discrete cosine transformation or Hadamard transformation can be used. If a square block consisting of n×n pixels is used, the plurality of transform coefficients will also be n×n per block.

Then, the transform coefficients are quantized and a quantization index for each transform coefficient is output.

FIG. 1 is a block diagram showing an embodiment of an image signal encoding device of the present invention. Note that in the following description, a two-dimensional discrete cosine transform (DCT) is used as the orthogonal transform. or,
The CB component signal and the CR component signal are thinned out by half in the horizontal direction.
One block consists of 8 pixels horizontally, a total of 64 pixels, and 3 DCT
A conversion section shall be used. Further, the quantization index is assumed to be variable length encoded.

The image encoding device shown in FIG. 1 converts the image signal into a Y component signal, a C
a color conversion unit 1 that converts the B component signal and the CR component signal;
A thinning section 3 that thins out the B component signal to half in the horizontal direction, a thinning section 5 that thins out the CR component signal to half in the horizontal direction, and a Y component signal storage section that stores two blocks of Y component signals. 2, a CB component signal storage section 4 that stores one bronch of the thinned out CB component signal, and one bronch of the thinned out CR component signal.
A CR component signal storage unit 6 stores blocks, and three DCT transform units 8,910 perform discrete cosine transform on each of the Y component signal, CB component signal, and CR component signal in block units to obtain 8×8 transform coefficients. , a quantization unit 11 that quantizes the transform coefficient and outputs a quantization index, and a variable length encoder 1 that performs variable length encoding on the quantization index.
2, and a Y component signal and a CB component signal so that there is no vacancy in the conversion processing operations in the three DCT conversion units 8, 9, and 10.

The CR component signal is stored in the storage unit 2 in block units. 4. 6, and distributes the data to three DCT transform units 8, 9, and 10, and transfers the transform coefficients from the DCT transform unit to the quantization unit 11.

Next, the operation of this embodiment will be explained with reference to FIG. 2. As described above, FIG. 2 is a diagram showing the order of blocks for reading out the Y component signal, the thinned out CB component signal, and the thinned out CR component signal. The RGB color image component signals 101 are converted into Y component signals 10 by the color converter 1.
2, a CB component signal 103, and a CR component signal 104. The Y component signal 102 is stored in the Y component signal storage section 2. The CB component signal thinning unit 3 extracts the CB component signal 10.
3 is horizontally thinned out to 1/2. The thinned out CB component signal 105 is stored in the CB component signal storage section 4.

The CR component signal thinning section 5 thins out the CR component signal 104 by half in the horizontal direction. CR component signal 1 after thinning
06 is stored in the CR component signal storage section 6.

While one pixel is stored in the CB component signal storage section 4 and the CR component signal storage section 6, two pixels are stored in the Y component signal storage section 2. Color conversion section 1, CB component signal thinning section 3
And the CR component signal thinning unit 5 repeats the above processing until the end of the image.

First, the control unit 7 monitors the storage capacity of the Y component signal storage unit 2 and the CB component signal storage unit 4, and determines whether the storage capacity of the Y component signal storage unit 2 is equal to 2 blocks and the CB component signal storage unit When the storage capacity of 4 becomes one block, the Y component signal 102 of 64 pixels corresponding to one block of the first block 21 is read out and sent to the DCT conversion section 8. At the same time, the control unit 7 controls the Y component signal 1 of 64 pixels for one block of the second block 22.
02 is read out and sent to the DCT conversion unit 9. At the same time, the control unit 7 controls the C of 64 pixels for one block of the first block 23.
The B component signal 103 is read out and sent to the DCT conversion section 10. This completes the first reading process indicated by block 33 in FIG.

The DCT transform unit 8 performs a two-dimensional discrete cosine transform on the Y component signal 102 for one block, and calculates and stores 8×8 transform coefficients 107. When the DCT conversion unit 8 completes the calculation of the input block, it notifies the control unit 7 that the calculation has ended.

The DCT transform unit 9 performs a two-dimensional discrete cosine transform on the sent Y component signal 102 for one block, and calculates and stores eight transform coefficients 108. When the DCT conversion unit 9 completes the calculation of the input block, it notifies the control unit 7 that the calculation has been completed.

The DCT transform unit 10 performs two-dimensional discrete cosine transform on the sent one block worth of CB component signal 105, calculates and stores eight transform coefficients 109. When the DCT conversion unit 10 completes the calculation of the input block, it notifies the control unit 7 that the calculation has ended.

Next, the control unit 7 monitors the states of the quantization unit 11 and the DCT conversion unit 8, and determines whether the processing of the quantization unit 11 is completed and the
After waiting for the calculation of the CT transformer 8 to be completed, the transform coefficients 107 of the DCT transformer 8 are read out and sent to the quantizer 11.

The quantizer 11 quantizes the received transform coefficients and sends the quantization index 110 to the variable length encoder 12.

When the quantization unit 11 completes the processing for one block, it notifies the control unit 7 that the processing has ended.

The variable length encoding unit 12 encodes and outputs the received quantization index 110. The variable length encoding unit 12
The above process is repeated until the end of the image.

Next, the control unit 7 monitors the states of the quantization unit 11 and the DCT conversion unit 9, and determines whether the processing of the quantization unit 11 is completed and the
After waiting for the calculation of the CT transformer 9 to be completed, the transform coefficients 108 of the DCT transformer 9 are read out and sent to the quantizer 11.

The quantizer 11 quantizes the received transform coefficients and sends the quantization index 110 to the variable length encoder 12.

When the quantization unit 11 completes the processing for one block, it notifies the control unit 7 that the processing has ended.

Next, the control unit 7 controls the quantization unit 11 and the DCT conversion unit 10.
monitors the state of , the processing of the quantization unit 11 is completed, and
After waiting for the calculation of the DCT transformer 10 to be completed, the transform coefficients 109 of the DCT transformer 10 are read out and sent to the quantizer 11.

The quantization unit 11 quantizes the received transform coefficients and sends a quantization index 110 to the variable length encoding unit 12. When the processing for one block is completed, the controller 7 is notified that the processing has been completed.

Next, the control unit 7 monitors the storage capacity of the CR component signal storage unit 6 and the Y component signal storage unit 2, and
When the storage capacity of the section 6 reaches one block and the storage capacity of the Y component signal storage section 2 reaches two blocks, the CR component signal 106 of 64 pixels corresponding to one block of the first block 24 is read out. , and sent to the DCT converter 8. At the same time, the control unit 7 reads out the Y component signal 102 of 64 pixels for one block of the third processor 725 and sends it to the DCT conversion unit 9. At the same time, the control unit 7 reads out the Y component signal 102 of 64 pixels for one block of the fourth block 26 and sends it to the DCT conversion unit 10.

This completes the second read process indicated by block 34 in FIG.

The DCT transform unit 8 performs a two-dimensional discrete cosine transform on the CR component signal 106 for one block that is sent, and calculates and stores eight transform coefficients 107. When the DCT conversion unit 8 completes the manual calculation of the block, it notifies the control unit 7 that the calculation has ended.

The DCT transform unit 9 performs a two-dimensional discrete cosine transform on the sent Y component signal 102 for one block, and calculates and stores eight transform coefficients 108. When the DCT conversion unit 9 completes the calculation of the input block, it notifies the control unit 7 that the calculation has ended.

The DCT transform unit 10 performs a two-dimensional discrete cosine transform on the Y component signal 102 for one block that has been sent, and calculates and stores eight transform coefficients 109. The DCT conversion unit 10
When the calculation of the input block is completed, the controller 7 is notified that the calculation has been completed.

Next, the control unit 7 monitors the states of the quantization unit 11 and the DCT conversion unit 8, and determines whether the processing of the quantization unit 11 is completed and the
After waiting for the calculation of the CT transformer 8 to be completed, the transform coefficients 107 of the DCT transformer 8 are read out and sent to the quantizer 11.

The quantizer 11 quantizes the received transform coefficients and sends the quantization index 110 to the variable length encoder 12.

(2) When the processing for a block is completed, the controller 7 is notified that the processing has been completed.

Next, the control unit 7 monitors the states of the quantization unit 11 and the DCT conversion unit 9, and determines whether the processing of the quantization unit 11 is completed and the
After waiting for the calculation of the CT transformer 9 to be completed, the transform coefficients 108 of the DCT transformer 9 are read out and sent to the quantizer 11.

The quantizer 11 quantizes the received transform coefficients and sends the quantization index 110 to the variable length encoder 12.

When the quantization unit 11 completes the processing for one block, it notifies the control unit 7 that the processing has ended.

Next, the control section 7, the quantization section 11 and the DCT conversion section 10
monitors the state of , the processing of the quantization unit 11 is completed, and
After waiting for the calculation of the DCT transformer 10 to be completed, the transform coefficients 109 of the DCT transformer 10 are read out and sent to the quantizer 11.

The quantizer 11 quantizes the received transform coefficients and sends the quantization index 110 to the variable length encoder 12.

When the quantization unit 11 completes the processing for one block, it notifies the control unit 7 that the processing has ended.Next, the control unit 7
The storage capacity of the CB component signal storage section 4 and the storage capacity of the CR component signal storage section 6 and Y component signal storage section 2 are monitored, and the CB
When the storage capacity of the component signal storage section 4 becomes one block, the storage capacity of the CR component signal storage section 6 becomes one block, and the storage capacity of the Y component signal storage section 2 becomes one block. , the CB component signal 105 of 64 pixels for one block of the second block 27 is read out and sent to the DCT conversion unit 8. At the same time, the control unit 7 reads out the CR component signal 106 of 64 pixels for one block of the second block 28, and
It is sent to the T converter 9. At the same time, the control section 7 reads out the Y component signal 102 of 64 pixels for one block of the fifth block 29 and sends it to the DCT conversion section 10. This completes the third read process indicated by block 35 in FIG.

The DCT transform unit 8 performs two-dimensional discrete cosine transform on the sent one block worth of CB component signal 105, calculates and stores 8×8 transform coefficients 107. When the DCT conversion unit 8 completes the calculation of the input block, it notifies the control unit 7 that the calculation has ended.

The DCT transform unit 9 performs two-dimensional discrete cosine transform on the sent CR component signal 106 for one block, calculates and stores 8×8 transform coefficients 108. When the DCT conversion unit 9 completes the calculation of the input block, it notifies the control unit 7 that the calculation has ended.

The DCT transform unit 10 performs two-dimensional discrete cosine transform on the Y component signal 102 for one block that has been sent, and calculates and stores 8×8 transform coefficients 109. The DCT conversion unit 10
When the calculation of the input block is completed, the controller 7 is notified that the calculation has been completed.

Next, the control unit 7 monitors the states of the quantization unit 11 and the DCT conversion unit 8, and determines whether the processing of the quantization unit 11 is completed and the
After waiting for the calculation of the CT transformer 8 to be completed, the transform coefficients 107 of the DCT transformer 8 are read out and sent to the quantizer 11.

The quantizer 11 quantizes the received transform coefficients and sends the quantization index 110 to the variable length encoder 12.

When the quantization unit 11 completes the processing for one block, it notifies the control unit 7 that the processing has ended.

Next, the control unit 7 monitors the states of the quantization unit 11 and the DCT conversion unit 9, and determines whether the processing of the quantization unit 11 is completed and the
After waiting for the calculation of the CT transformer 9 to be completed, the transform coefficients 108 of the DCT transformer 9 are read out and sent to the quantizer 11.

The quantizer 11 quantizes the received transform coefficients and sends the quantization index 110 to the variable length encoder 12.

When the quantization unit 11 completes the processing for one block, it notifies the control unit 7 that the processing has ended.

Next, the control unit 7 controls the quantization unit 11 and the DCT conversion unit 10.
monitors the state of , the processing of the quantization unit 11 is completed, and
After waiting for the calculation of the DCT transformer 10 to be completed, the transform coefficients 109 of the DCT transformer 10 are read out and sent to the quantizer 11.

The quantizer 11 quantizes the received transform coefficients and sends the quantization index 110 to the variable length encoder 12.

When the processing for one block is completed, the controller 7 is notified that the processing has been completed.

The control unit 7 repeats the above processing until the end of the image.

In the above description, it is assumed that the number of pixels is thinned out by half in the horizontal direction, but the thinning process may be performed in both the horizontal and vertical directions.

In addition, in the above explanation, the R component signal, the G component signal,
Although the B component signal is color-converted into a Y component signal, a CB component signal, and a CR component signal, other component signals such as Y
The color may be converted into a component signal, a (RY) component signal, or a (B-Y) component signal.

Also, from the Y component signal, CB component signal, and CR component signal,
It is also effective when performing color conversion into component signals, G component signals, B component signals, etc.

〔Effect of the invention〕

As described above, by using the image signal encoding device of the present invention, it is possible to effectively use a plurality of orthogonal transform units and perform high-speed encoding.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an embodiment of the image signal encoding device of the present invention, and FIG. 2 shows the order of blocks for reading out the Y component signal, the thinned out CB component signal, and the thinned out CR component signal. FIG. l...Color conversion unit 2...Y component signal storage unit 3...CB component signal thinning unit 4...CB component signal storage unit 5...CR acid Component signal spacing unit 6...CR component signal storage unit 7...Control unit 8.9.10...DCT conversion unit 11...Quantization unit 12... Variable length encoder 21...First block 22 of Y component signal...
- Second block 23 of the Y component signal...First block 24 of the thinned out CB component signal...First block 25 of the thinned out CR component signal...Y component Third block of signals 26...
・Fourth block 27 of the Y component signal...Second block 28 of the thinned out CB component signal...Second block 29 of the thinned out CR component signal...Y component Fifth block of signals 30...
- Sixth block 31 of the Y component signal...Third block 32 of the thinned out CB component signal...Third block of the thinned out CR component signal

Claims (1)

[Claims] (1) a color conversion unit that converts an image signal into a plurality of component signals; a thinning unit that thins out at least one component signal among the plurality of color-converted component signals; and the component signal that has not been thinned out and the a storage unit that stores the thinned out component signals for a plurality of pixels; and a storage unit that performs orthogonal transformation on the unthinned component signals and the thinned out component signals in units of blocks each consisting of a plurality of pixels, and generates a plurality of transform coefficients. A quantization unit that quantizes the transform coefficients and outputs a quantization index; An encoding unit that encodes the quantization index; and a transform processing operation in the orthogonal transform units. The unthinned component signals and the thinned out component signals are read out from the storage unit in blocks each consisting of a plurality of pixels and distributed to the plurality of orthogonal transformation units so that there is no empty space in the orthogonal transformation unit. An image encoding device comprising: a control section that transfers the transform coefficients from the transform section to the quantization section.