JP2000013793A - Image encoding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image encoding device capable of accurately controlling the amount of codes after compression encoding while reducing circuit scale. SOLUTION: A code amount control part 17 outputs a quantization step control signal SC and mask control signal SM corresponding to a generated code amount CM. A quantizing part 20 sets the value of a quantizing step from a memory 25 corresponding to the quantizing step control signal SC. Besides, when mask processing is instructed by the mask control signal SM, in place of the value set from the memory 25, a prescribed value, which is stored in a storage means 24b, to extremely reduce the code amount at the time of performing variable length encoding to output data from the quantizing part 20 is forcedly set as the quantizing step by a mask processing part 24. Thus, even when data having the large amount of codes is continued, for example, by instructing mask processing through the mask control signal SM, the code amount can be accurately controlled. Besides, the mask processing part 24 can be extremely simply composed of a selector 24a and storage means 24b.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image encoding apparatus for compressing and encoding a digitized image signal, and more particularly to a technique for controlling the code amount of compression-encoded data.

[0002]

2. Description of the Related Art In recording a digitized image signal, it is necessary to compress and record an enormous amount of information in order to reduce the size of a recording / reproducing apparatus. For this,
Motion compensated prediction, discrete cosine transform (Discrete Cosine Tra)
nsform: DCT) and variable-length coding (Variable Length Codin) such as Huffman coding in addition to band division.
g: VLC), compresses the digital image signal by arithmetic coding, etc., and transmits and stores it.

For example, MPE which is an international standard for moving picture compression
In G, the compression ratio of an image signal is improved by performing processing based on a unit called a macroblock composed of 16 pixels in the horizontal direction and 16 pixels in the horizontal direction.

FIG. 8 is a block diagram showing a configuration example of a conventional image coding apparatus. First, macroblock data is input as an input image 1. ME unit 51 that performs motion detection
Finds the one that most closely matches the input image 1 in the 16 × 16 area of the data stored in the frame memory 52. Then, the position is output as a motion vector to the predicted image generation unit 53 that generates a predicted image. Predicted image generation unit 53
Generates a predicted image from the frame memory 54 based on the input motion vector. Then, according to the difference between the predicted image and the input image 1, the difference between the input image 1 and the predicted image or the input image 1 itself is input to the DCT unit 55 as a difference image.

[0005] A DCT unit 55 performs a discrete cosine transform (DCT) on the difference image and outputs the result. The quantization unit 60 performs quantization on the output of the DCT unit 55,
The result is output as quantized data. The variable-length coding unit 56 performs variable-length coding on the quantized data and
Output as At the same time, the code amount of the code data 2 is output to the code amount control unit 57 as the generated code amount CM.

On the other hand, an inverse quantization unit 58 inversely quantizes the quantized data and outputs the result to the IDCT unit 59 as a DCT transform coefficient. The IDCT unit 59 performs an inverse discrete cosine transform (IDCT) on the output of the inverse quantization unit 58 and outputs a local reproduced image LC. The local reproduction image LC is stored in the frame memory 54, and is used by the predicted image generation unit 53 to perform motion compensation.

The code amount control section 57 outputs a quantization step control signal SC to the quantization section 60 based on the generated code amount CM output from the variable length coding section 56 to perform quantization. Is changed.

[0008] In such a series of encoding processes, the reason for changing the quantization step when performing quantization is that variable length coding reduces the code length stochastically. This is because the code amount does not always decrease. According to the MPEG standard, the quantization step can be changed in the range of 1 to 31 in order to control the code generation amount.

[0009]

In order to control the code amount with high accuracy, an accurate generated code amount is required. However, in the system as shown in FIG. 8, the accurate generated code amount can be known only after performing variable length coding. Therefore, when the generated code amount is too large as a result of the variable length coding, control for reducing the code amount is required.

As a method of controlling the code generation amount after performing variable length coding, a method of controlling the generated code amount using a matrix is known (see Japanese Patent Application Laid-Open No. 9-23427). However, in this method, there is a problem that the circuit scale becomes large because it is necessary to have an extra control matrix.

In view of the above problems, an object of the present invention is to provide an image coding apparatus which can control the amount of code after compression coding with high accuracy and has a small circuit scale.

[0012]

According to a first aspect of the present invention, there is provided an image coding apparatus for compressing and encoding a digitized image signal by blocking the image signal into a plurality of pixels. A conversion unit that converts an image signal of each block into a plurality of coefficients, a quantization unit that quantizes the plurality of coefficients using a quantization step, and compression-encodes output data of the quantization unit. A compression step for controlling a value of a quantization step in the quantization section according to a code amount of data compression-coded by the compression coding section; And a code amount control unit that outputs a mask control signal that instructs whether or not to perform mask processing on the quantization unit, wherein the quantization unit sets a value of a quantization step according to the quantization step control signal. Amount to do When the execution of the mask processing is instructed by the quantization step setting means and the mask control signal, instead of the value set by the quantization step setting means, the output data of the quantization unit is compression-encoded. And a mask processing unit for forcibly setting a predetermined value such that the code amount of the data is extremely small as a value of the quantization step.

According to the first aspect of the present invention, the code amount control section outputs a quantization step control signal and a mask control signal according to the code amount of the compression-encoded data. The quantization unit sets a value of a quantization step by a quantization step setting unit according to the quantization step control signal. In addition, when the execution of the mask processing is instructed by the mask control signal, the output data of the quantization unit is replaced by the mask processing unit instead of the value set by the quantization step setting unit. A predetermined value such that the amount of code at the time of compression encoding becomes extremely small is compulsorily set as the value of the quantization step. Thus, for example, even when data having a large code amount continues, the code amount can be accurately controlled by instructing execution of the mask process by the mask control signal. Further, since the mask processing unit sets a predetermined value as the value of the quantization step, the circuit scale can be extremely small.

According to a second aspect of the present invention, in the image encoding apparatus of the first aspect, the mask processing unit is configured to store the predetermined value and to instruct execution of the mask processing by the mask control signal. In the case where the predetermined value stored in the storage means is selected and output as the value of the quantization step, otherwise, the value set by the quantization step setting means is selected as the value of the quantization step. And an output selector.

According to the third aspect of the present invention, in the first aspect,
The code amount control unit in the image coding apparatus of (1), when the code amount of the compression-encoded data exceeds a predetermined value, instructs execution of mask processing by the mask control signal.

[0016]

FIG. 1 is a block diagram showing a configuration of an image coding apparatus according to one embodiment of the present invention. The image encoding device shown in FIG. 1 operates as follows.

When the input image 1 is input, the ME unit 11 for performing motion detection performs a
The one that best matches the input image 1 is obtained from the 6 × 16 area. Then, the position of the best matching data is output as a motion vector to the predicted image generation unit 13 that generates a predicted image.

The predicted image generator 13 generates a predicted image from the data stored in the frame memory 14 based on the input motion vector. Then, according to the difference between the predicted image and the input image 1, the difference between the input image 1 and the predicted image or the input image 1 itself is input to the DCT unit 15 as a difference image.

The DCT unit 15 as a transform unit performs a discrete cosine transform (DCT) on the difference image, and outputs the result as a DCT transform coefficient DCT as a plurality of coefficients. The quantization unit 20 performs quantization and mask processing on the output of the DCT unit 15 and outputs the result as quantized data.

Variable length coding section 16 as compression coding section
Variable-length-encodes the quantized data output from the quantization unit 20 and outputs it as code data 2. With this,
The code amount of the generated code is output to the code amount control unit 17 as a generated code amount CM.

On the other hand, the inverse quantization unit 18 inversely quantizes the quantized data output from the quantization unit 20 and outputs the result to the IDCT unit 19 as a DCT transform coefficient. The IDCT unit 19 applies an inverse discrete cosine transform (ID
CT) to output a local reproduced image LC. The local reproduction image LC is stored in the frame memory 14, and is used by the predicted image generation unit 13 to perform motion compensation.

The code amount control unit 17 outputs a quantization step control signal SC to the quantization unit 20 based on the generated code amount CM output from the variable length coding unit 16 to perform quantization. The number of generated steps is controlled by changing the number of quantization steps. When the generated code amount CM exceeds the predetermined code amount, the quantization unit 20 is instructed by the mask control signal SM to perform a process of masking the DCT transform coefficient DCT output from the DCT unit 15.

Briefly, quantization in MPEG means that DCT transform coefficients are firstly divided by a coefficient called a quantization matrix, and then divided by a coefficient called the number of quantization steps. .
In the present embodiment, the quantization unit 20 outputs the mask control signal SM
When the mask processing is instructed, a predetermined value that minimizes the code amount when the output data of the quantization unit 20 is compression-encoded is forcibly set as the quantization step instead of the quantization step. Set and perform division.

FIG. 2 is a block diagram showing an example of the internal configuration of the quantization section 20 shown in FIG. In FIG. 2, reference numeral 24 denotes a mask processing unit, which controls whether or not to execute a mask process by a mask control signal SM from the code amount control unit 17.

The quantization section 20 shown in FIG. 2 operates as follows.

The quantization unit 20 receives as input the DCT transform coefficient DCT output from the DCT unit 15. The input data is macroblock data as shown in FIG. The macro block is composed of four luminance component blocks Y0, Y1, Y2, and Y3 of 8 × 8 pixels and two chroma component blocks Cb and Cr of 8 × 8 pixels.

The data of each of the blocks Y0 to Y3, Cb, and Cr constituting the macroblock is input in pixel units in the order shown in FIG. The numbers and arrows in each frame in FIG. 4 indicate the input order.

The divider 21 divides each input pixel by the quantization matrix stored in the memory 22. FIG. 5 is a diagram illustrating an example of the quantization matrix. The quantization matrix shown in FIG. 5 has, as a component, a value corresponding to each pixel of the block of 8 × 8 pixels, and the divider 21 uses the value of the component of the quantization matrix corresponding to the input pixel. , Perform division.

The result of the division by the divider 21 is input to the divider 23 as an intermediate division result MD. The divider 23 divides the intermediate division result MD by the number of quantization steps output from the mask processing unit 24. The mask processing unit 24 selectively outputs the number of quantization steps stored in the memory 25 as the quantization step setting means or a predetermined value stored in the storage means 24b by the selector 24a. Switching of the selector 24a is controlled by the mask control signal SM. The division result RD by the divider 23 is output to the variable length coding unit 16 and the inverse quantization unit 18 as quantized data.

FIG. 6 is a diagram showing an example of the number of quantization steps stored in the memory 25. In MPEG, the number of quantization steps can be changed in the range of 1 to 31, and the number of quantization steps corresponding to each quantization step is set in the memory 25.

The selector 24a of the mask processing unit 24 is controlled to be switched, for example, as follows.

The variable-length encoding unit 16 having received the division result RD performs variable-length encoding and outputs the code data 2, calculates the code amount at this time, and calculates the code amount at this time as the generated code amount CM. Output to The code amount control unit 17 controls the quantization in the quantization unit 20 based on the received generated code amount CM. Here, assuming that the ratio of the generated code amount to the maximum generated code amount is X (%) and the quantization step is N, the relationship between the ratio X and the quantization step N can be set as follows. 0 ≦ X <10 → N = 110 ≦ X <20 → N = 2 20 ≦ X <30 → N = 4 30 ≦ X <40 → N = 640 <X <50 → N = 950 <X <60 → N = 12 60 ≦ X <70 → N = 16 70 ≦ X <80 → N = 20 80 ≦ X <90 → N = 25 90 ≦ X ≦ 100 → N = 31 The code amount control unit 17 is, for example, Based on such a relationship, the memory 25
Is selected from the number of quantization steps stored in. At the same time, the mask control signal SM
The selector 24 a is controlled by the selector 23, and the number of quantization steps from the memory 25 is input to the divider 23. Normally,
The generated code amount is controlled by such processing.

However, the quantization step can take only values from 1 to 31 according to the MPEG standard. For this reason, for example, when a complex image in which hundreds of macroblocks having a ratio X of 80% or more continue over several hundreds, a buffer for temporarily holding the code data 2 output from the variable length coding unit 16 The memory may overflow.

Therefore, when the generated code amount cannot be controlled only by the quantization step, the code amount control unit 1
7 switches and controls the selector 24a by the mask control signal SM, and inputs the predetermined value stored in the storage means 24b to the divider 23.

Here, if a very large value, for example, 10,000 is set in the storage means 24b as the predetermined value in a range that the intermediate division result MD can take, the division result RD becomes a value less than ± 0.5. become. According to the MPEG standard, the input of the variable-length encoding unit 16 must be an integer, so that when the division result RD is rounded to an integer, it becomes “0”. The variable-length encoding unit 16 encodes all "0" s consecutively, so that the generated code amount can be reliably reduced.

However, if such processing is performed uniformly for all blocks, the division results RD all become "0", so that the image quality is expected to be significantly deteriorated.
For this reason, mask processing is performed only on a certain specific block in the macro block, or mask processing is performed only on a specific pixel of each block (for example, the 29th pixel and the following pixels in FIG. 4). ,
It is also possible to minimize the effect on image quality.
Further, the predetermined value may have a plurality of values.

The quantizer according to the present embodiment can be configured using a multiplier instead of a divider. FIG. 7 is a block diagram showing another example of the internal configuration of the quantization unit shown in FIG. 1. The quantization using multipliers 31 and 33 in place of the dividers 21 and 23 in the quantization unit 20 shown in FIG. It is a figure showing part 20A. The quantization unit 20A shown in FIG.
Then, the reciprocal of each value of the quantization matrix stored in the memory 22 in FIG. 2 is stored in the memory 32, and the reciprocal of each value of the quantization step number stored in the memory 25 in FIG. It should be done. Of course, the quantization unit may be configured such that one of the operation of the quantization matrix and the operation of the quantization step is performed by the divider, and the other is performed by the multiplier.

[0038]

As described above, according to the present invention, even when, for example, data having a large code amount continues, the code amount can be accurately controlled by designating a mask with a mask control signal. . Further, since the mask processing section sets a predetermined value as the quantization step, the circuit scale can be extremely small. Therefore, it is possible to control the amount of code with high accuracy and to realize an image coding device with a small circuit scale.

[Brief description of the drawings]

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

FIG. 2 is a diagram illustrating an example of an internal configuration of a quantization unit illustrated in FIG. 1;

FIG. 3 is a diagram illustrating an example of a DCT transform coefficient according to an embodiment of the present invention.

FIG. 4 is a diagram illustrating an example of an input order of DCT transform coefficients to a quantization unit according to an embodiment of the present invention.

FIG. 5 is a diagram illustrating an example of a quantization matrix according to an embodiment of the present invention.

FIG. 6 is a diagram illustrating an example of the number of quantization steps according to an embodiment of the present invention.

FIG. 7 is a diagram illustrating another example of the internal configuration of the quantization unit illustrated in FIG. 1;

FIG. 8 is a block diagram illustrating a configuration of a conventional image encoding device.

[Explanation of symbols]

 Reference Signs List 15 DCT section (transformation section) 20, 20A Quantization section 16 Variable length coding section (compression coding section) 24 Mask processing section 24a Selector 24b Storage means 25 Memory (quantization step setting means) SC Quantization step control signal SM Mask control signal DCT DCT transform coefficient (multiple coefficients)

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Shuji Inoue 1006 Kadoma Kadoma, Osaka Prefecture Inside Matsushita Electric Industrial Co., Ltd. Terms (reference) 5C059 KK22 MA00 MA23 MC11 PP04 TA46 TC02 TC06 TC18 TD12 UA02 UA31 5C078 AA04 BA57 CA01 DA01 DB07

Claims (3)

[Claims] An image coding apparatus for compressing and encoding a digitized image signal into blocks for each of a plurality of pixels, wherein the conversion unit converts the image signal for each block into a plurality of coefficients. A quantization unit that quantizes a plurality of coefficients using a quantization step; a compression encoding unit that compresses and encodes output data of the quantization unit; and a code of data that is compression encoded by the compression encoding unit. A code for outputting a quantization step control signal for controlling a value of a quantization step in the quantization unit according to the amount, and a mask control signal for instructing whether or not to perform a mask process on the quantization unit. A quantization step setting means for setting a value of a quantization step in accordance with the quantization step control signal; and a mask processing signal by the mask control signal. Is designated, instead of the value set by the quantization step setting means, a predetermined value such that the code amount when the output data of the quantization unit is compression-encoded is extremely small, An image coding apparatus comprising: a mask processing unit forcibly setting a value of an encoding step. 2. The image encoding device according to claim 1, wherein the mask processing unit is configured to: store the predetermined value; and when the execution of the mask processing is instructed by the mask control signal, A selector for selectively outputting the predetermined value stored in the storage means as the value of the quantization step, and otherwise selecting and outputting the value set by the quantization step setting means as the value of the quantization step. An image encoding device, comprising: 3. The image encoding apparatus according to claim 1, wherein the code amount control unit executes a mask process by the mask control signal when a code amount of the data compressed and coded exceeds a predetermined value. An image coding apparatus for instructing image coding.