JP2006229650A - Device, method, and program for encoding image - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To encode an image, while relieving burden on discrete cosine transformation and quantizing processing. <P>SOLUTION: When a difference absolute value sum calculating part 5 calculates a difference absolute value sum SS, a skip processing part 1 judges whether the discrete cosine transformation and quantization are skipped or not at each block, based on a comparison result with a quantizing parameter QM to be used for the quantization. Then skip processing is performed concerning the discrete cosine transformation and the quantization concerning each block. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image encoding device, an image encoding method, and an image encoding program, and is particularly suitable for application to a discrete cosine transform process and a quantization process skip method in image processing.

  In recent years, with the spread of the Internet, mobile phones, and the like, compressed images are transmitted and received. Here, as methods for compressing images, there are standards such as JPEG and MPEG-4. In these methods, image compression is performed by performing discrete cosine transform (Discrete Cosine Transform) and quantization processing of image data. The rate is increasing.

Further, for example, Patent Document 1 discloses a method for reducing the processing load as a whole by skipping division in quantization processing in a predetermined case.
In these methods, before the quantization process, it is individually determined whether or not each DCT coefficient included in the block is “0” in the quantization process, and when it is determined to be “0”, Skip the quantization process.

For example, when “x” is a DCT coefficient to be quantized, “K” is a divisor based on a quantization parameter, and an operation in quantization is performed by x / K, “y” obtained by y = x / K If −K <| x | <K, it can be determined that y = 0 without performing division. In this case, the division result is rounded down.
JP-A-10-191334

However, in the method disclosed in Patent Document 1, in order to determine whether or not each DCT coefficient becomes “0” in the quantization process, it is necessary to calculate the DCT coefficient for each block in advance. There is a problem that the burden on the cosine transform and the quantization process is large.
Accordingly, an object of the present invention is to provide an image encoding device, an image encoding method, and an image encoding program capable of performing image encoding while reducing the burden on discrete cosine transform and quantization processing. It is.

  In order to solve the above-described problem, according to an image encoding device according to an aspect of the present invention, a discrete cosine transform processing unit that performs discrete cosine transform for each block including a plurality of pixels obtained by dividing an image. A quantization unit that performs quantization on the discrete cosine transform result obtained by the discrete cosine transform processing unit, an average value of pixel values of each block before the discrete cosine transform, and the quantization And a skip processing unit that performs discrete cosine transform and quantization skip processing on the block based on a comparison result with a function including a quantization parameter to be used as a variable.

  As a result, it is possible to determine whether the quantization result after the discrete cosine transform becomes zero before performing the discrete cosine transform for each block, and without performing the discrete cosine transform, the discrete cosine transform and the quantization can be performed. Unnecessary blocks can be identified. Therefore, the number of blocks that need to be subjected to discrete cosine transform and quantization processing can be reduced, and the image encoding processing can be speeded up without degrading the image quality.

  Further, according to the image coding device according to an aspect of the present invention, the skip processing unit includes an average value of pixel values of each block before the discrete cosine transform, and a quantization parameter used for the quantization Based on the result of comparison with a function including a variable, zero is output as the discrete cosine transform and quantization result for the block without performing the discrete cosine transform and quantization for the block.

Thereby, the discrete cosine transform and the quantization result can be obtained without performing the discrete cosine transform and the quantization, and the burden on the discrete cosine transform and the quantization process can be reduced.
In addition, according to the image coding apparatus according to an aspect of the present invention, a discrete cosine transform processing unit that performs discrete cosine transform for each block including a plurality of pixels obtained by dividing an image, and the discrete cosine transform processing A quantization unit that performs quantization on the discrete cosine transform result obtained in the unit, and a sum of absolute values of differences between pixel values and an average value of pixel values of each block before the discrete cosine transform, And a skip processing unit that performs discrete cosine transform and quantization skip processing on the block based on a comparison result with a function including a quantization parameter used as a variable as a variable.

  Thereby, it is possible to determine whether or not all of the AC coefficients except for the DC coefficient of the DCT coefficient are zero before performing the discrete cosine transform for each block. For this reason, it is possible to specify blocks that do not require discrete cosine transform and quantization without performing discrete cosine transform, and determine whether all of the DCT coefficients including the DC coefficient are zero. As compared with, the number of blocks that do not require discrete cosine transform and quantization can be increased. As a result, the number of blocks that need to be subjected to discrete cosine transform and quantization processing can be further reduced, and the image encoding processing can be further speeded up without degrading the image quality.

  Further, according to the image coding device according to an aspect of the present invention, the skip processing unit includes an absolute difference value between an average value of pixel values of each block and each pixel value before the discrete cosine transform. Based on the comparison result between the sum and a function including the quantization parameter used for the quantization as a variable, the discrete cosine transform and quantization for the block without performing the discrete cosine transform and quantization for the block As a result, zero is output.

Thereby, the discrete cosine transform and the quantization result can be obtained without performing the discrete cosine transform and the quantization, and the burden on the discrete cosine transform and the quantization process can be reduced.
In addition, according to the image encoding device according to an aspect of the present invention, a discrete cosine transform processing unit and a processor that performs quantization processing for each block including a plurality of pixels obtained by dividing an image, and the image A first memory for storing the image data, a second memory incorporated in the processor, a DMA controller for DMA transfer of the image data stored in the first memory to the second memory, and at the time of the DMA transfer A pixel sum calculation unit that calculates a sum of pixel values of each block before being subjected to the discrete cosine transform, and the processor based on the sum of the pixel values calculated by the pixel sum calculation unit, The sum of absolute differences between the average value of the pixel values of each block before the discrete cosine transform and each pixel value is calculated, and is used for the calculated absolute difference sum and the quantization. That on the basis of a comparison result of a function including the quantization parameters as variables, and performs skip processing of the discrete cosine transform and quantization of the block.

  This allows the block other than the processor to calculate the sum of the pixel values of the block during DMA transfer. For this reason, it becomes possible to calculate the sum of absolute differences while suppressing the load on the processor, and before performing the discrete cosine transform for each block, all the AC coefficients except for the DC coefficient of the DCT coefficient are made zero. Can be determined. As a result, the number of blocks that need to be subjected to discrete cosine transform and quantization processing can be reduced, and the image encoding processing can be speeded up without degrading the image quality.

  In addition, according to the image coding method according to an aspect of the present invention, the sum of absolute differences between the average value of the pixel values of each block and each pixel value before the discrete cosine transform, and the quantization parameter Comparing a function including a variable, and if the sum of absolute differences for the block is greater than a function including the quantization parameter as a variable, performing a discrete cosine transform and quantization for the block; and And a step of skipping discrete cosine transform and quantization for the block when the sum of absolute differences for the block is smaller than a function including the quantization parameter as a variable.

  Thus, before performing the discrete cosine transform for each block, it can be determined whether all the AC coefficients except for the DC coefficient of the DCT coefficient are zero, and the discrete cosine transform and quantization for the specific block can be performed. It is possible to skip. Therefore, the number of blocks that need to be subjected to discrete cosine transform and quantization processing can be reduced, and the image encoding processing can be speeded up without degrading the image quality.

Further, according to the image coding method according to an aspect of the present invention, when the sum of absolute differences for the block is smaller than a function including the quantization parameter as a variable, discrete cosine transform and quantization for the block Without performing the above, zero is output as the discrete cosine transform and quantization result for the block.
Thereby, the discrete cosine transform and the quantization result can be obtained without performing the discrete cosine transform and the quantization, and the burden on the discrete cosine transform and the quantization process can be reduced.

  In addition, according to the image coding method according to an aspect of the present invention, the step of performing DMA transfer of the image data stored in the first memory to the second memory incorporated in the processor, and the discrete cosine at the time of the DMA transfer The step of causing the hardware to calculate the sum of the pixel values of each block before the conversion and the sum of the pixel values calculated by the hardware are used on the processor before the discrete cosine transform is performed. A step of calculating a sum of absolute differences between the average value of the pixel values of each block and each pixel value, and a function including the calculated sum of absolute differences of the blocks and a quantization parameter as variables. And when the difference absolute value sum for the block is larger than the function including the quantization parameter as a variable, A step of performing discrete cosine transform and quantization on the block on the processor while accessing image data stored in the second memory; and a sum of absolute differences of the block using the quantization parameter as a variable And a step of skipping discrete cosine transform and quantization for the block if smaller than the function to include.

  As a result, it is possible to calculate the sum of absolute differences while suppressing the load on the processor, and before performing the discrete cosine transform for each block, all the AC coefficients except for the DC coefficient of the DCT coefficient are set to zero. Can be determined. As a result, it is possible to skip discrete cosine transform and quantization for a specific block, and image coding can be performed while reducing the burden on discrete cosine transform and quantization processing.

  In addition, according to the image coding program according to the aspect of the present invention, the sum of absolute differences between the average value of the pixel values of each block before the discrete cosine transform and each pixel value, and the quantization parameter are calculated. Comparing a function including a variable, and if the sum of absolute differences for the block is greater than a function including the quantization parameter as a variable, performing a discrete cosine transform and quantization for the block; and When the sum of absolute differences for a block is smaller than a function including the quantization parameter as a variable, the computer is caused to execute a step of skipping discrete cosine transform and quantization for the block.

  Thus, by causing the computer to execute the image encoding program, it is possible to determine whether or not all of the AC coefficients except for the DC coefficient of the DCT coefficient are zero before performing the discrete cosine transform for each block. It becomes possible to skip discrete cosine transform and quantization for a specific block. For this reason, the number of blocks that need to be subjected to discrete cosine transform and quantization processing can be reduced while suppressing the increase in hardware scale, and image coding processing can be performed at high speed without degrading image quality. Can be realized.

According to the image coding program of one aspect of the present invention, when the sum of absolute differences for the block is smaller than a function including the quantization parameter as a variable, discrete cosine transform and quantization for the block Without performing the above, zero is output as the discrete cosine transform and quantization result for the block.
Thus, by causing the computer to execute the image encoding program, it is possible to obtain the discrete cosine transform and the quantization result without performing the discrete cosine transform and the quantization, while suppressing the increase in the scale of the hardware. The burden on the discrete cosine transform and the quantization process can be reduced.

  In addition, according to the image coding program of one aspect of the present invention, the step of performing DMA transfer of image data stored in the first memory to the second memory incorporated in the processor, and the discrete cosine at the time of the DMA transfer The step of causing the hardware to calculate the sum of the pixel values of each block before the conversion and the sum of the pixel values calculated by the hardware are used to calculate each block before the discrete cosine transform. A step of calculating a sum of absolute differences between an average value of pixel values and each pixel value, a step of comparing the calculated sum of absolute differences with respect to the block and a function including a quantization parameter as a variable, and the block If the sum of absolute differences is larger than the function including the quantization parameter as a variable, the image stored in the second memory Performing discrete cosine transform and quantization for the block while accessing the data, and if the sum of absolute differences for the block is smaller than the function including the quantization parameter as a variable, the discrete for the block And a step of skipping cosine transform and quantization.

  Thus, by causing the computer to execute the image encoding program, it is possible to calculate the sum of absolute differences while suppressing the load on the processor, and before performing the discrete cosine transform for each block, the DCT coefficient It can be determined whether all of the AC coefficients except for the DC coefficient are zero. As a result, it is possible to skip discrete cosine transform and quantization for a specific block, and image coding can be performed while reducing the burden on discrete cosine transform and quantization processing.

Hereinafter, an image encoding apparatus and method according to an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing a functional configuration of an image encoding device according to an embodiment of the present invention.
In FIG. 1, the image encoding apparatus includes a skip processing unit 1, a discrete cosine transform processing unit 2, a quantization unit 3, and an entropy encoding unit 4. The skip processing 1 includes a difference absolute value sum calculation unit 5. Is provided. In the following embodiment, a process targeted for intraframe encoding in still image encoding such as JPEG or moving image encoding such as MPEG-4 will be described.

  Here, the skip processing unit 1 is provided with pixel data of a frame to be encoded as input data. Then, the difference absolute value sum between the average pixel value of each block before each discrete cosine transform and each pixel value is calculated, and the difference absolute value sum is compared with a function including a quantization parameter as a variable. Based on the result, discrete cosine transform and quantization skip processing are performed for the block.

The discrete cosine transform processing unit 2 performs discrete cosine transform for each block including a plurality of pixels obtained by dividing a frame, and transforms image data of each block from the time domain to the spatial frequency domain.
Specifically, assuming that the function of pixel data is f (x, y), a function F (u, v) obtained by discrete cosine transform is given by the following equation.

  The quantization unit 3 quantizes the DCT coefficient calculated by the discrete cosine transform processing unit 2. Specifically, an operation including a process of dividing the DCT coefficient by a predetermined quantization parameter is performed, and a smaller value (quantization result) obtained by the quantization is set as the DCT coefficient. In this quantization process, it is possible to reduce the amount of encoding information by removing high-frequency components that do not affect human vision. For this reason, generally, the higher the coefficient of the quantization table, the larger the number can be set.

For example, in the JPEG method, when the quantization parameter is QM, the quantization result QF (u, v) of f (x, y) is given by the following equation.
If F (u, v) ≧ 0,
QF (u, v) = (F (u, v) −QM / 2) / QM
If F (u, v) <0,
QF (u, v) = (F (u, v) + QM / 2) / QM
... (2)
The entropy encoding unit 4 performs a variable length encoding process on the quantization result obtained by the quantization unit 3. That is, a short code is assigned to a symbol with a high appearance frequency (quantized DCT coefficient), and a long code is assigned to a symbol with a low appearance frequency (quantized DCT coefficient). Examples of entropy encoding include Huffman encoding and calculation encoding.

Then, when the pixel data of the frame to be encoded is given as input data to the skip processing unit 1, the difference absolute value sum calculation unit 5 calculates the average value of the pixel values of each block obtained by dividing the frame, and The sum of absolute differences between the pixel values is calculated.
Here, each block is composed of 8 × 8 = 64 pixels, and if the sum of the pixel values of the pixels constituting each block is p, the difference absolute value sum SS can be given by the following equation.

SS = | f (0,0) −p / 64 | + | f (0,1) −p / 64 | +...
+ | F (7,7) -p / 64 |
That is,

ただし、

However,

である。

It is.

  When the difference absolute value sum SS is calculated by the difference absolute value sum calculation unit 5, the skip processing unit 1 performs discrete processing for each block based on the comparison result with the quantization parameter QM used for quantization. It is possible to determine whether or not to perform cosine transform and quantization skip processing, and to perform discrete cosine transform and quantization skip processing for each block.

  That is, when the difference absolute value sum SS for a certain block is larger than the function including the quantization parameter QM as a variable, the skip processing unit 1 sends the image data for the block to the discrete cosine transform processing unit 2. When the image data for the block is sent to the discrete cosine transform processing unit 2, the discrete cosine transform processing unit 2 performs a discrete cosine transform for the block, and a quantization unit for the discrete cosine transform result for the block Send to 3. Then, when the discrete cosine transform result for the block is sent, the quantizing unit 3 quantizes the discrete cosine transform result for the block, and sends the quantized result to the entropy coding unit 4.

  On the other hand, when the difference absolute value sum SS for a certain block is smaller than the function including the quantization parameter QM as a variable, the skip processing unit 1 skips the discrete cosine transform and quantization for the block. That is, when the difference absolute value sum SS for a certain block is smaller than the function including the quantization parameter QM as a variable, the skip processing unit 1 sends the image data for the block to the discrete cosine transform processing unit 2. Instead, zero is sent to the entropy encoding unit 4 as a result of the discrete cosine transform and quantization for the block.

Specifically, when the sum of absolute differences SS for a block satisfies the following expression (5), all AC coefficients except for the DC coefficient of the DCT coefficient for the block are zero, and the discrete cosine for the block Transform and quantization can be skipped.
SS <6QM (5)
Here, when the sum of absolute differences SS satisfies the equation (5), it can be derived as follows that all the AC coefficients except for the DC coefficient of the DCT coefficient are zero.

  That is, the absolute value of the function F (u, v) obtained by the discrete cosine transform is as follows from the equation (1).

ここで、
|ｘ₁＋ｘ₂＋ｘ₃＋・・・≦|ｘ₁|＋|ｘ₂|＋|ｘ₃|・・・
が成り立つので、以下の（７）式が成り立つ。

here,
| x ₁ + x ₂ + x ₃ + ... ≦ | x ₁ | + | x ₂ | + | x ₃ | ...
Therefore, the following equation (7) is established.

また、
−１≦ｃｏｓθ≦＋１
であるので、以下の（８）式が成り立つ。

Also,
−1 ≦ cos θ ≦ + 1
Therefore, the following equation (8) is established.

また、Ｃ（ｕ），Ｃ（ｖ）は、

Also, C (u) and C (v) are

であることと、Ｎ＝８から、以下の（９）式が任意のｕ，ｖで成り立つ。

Since N = 8, the following equation (9) holds for arbitrary u and v.

一方、量子化結果ＱＦ（ｕ，ｖ）は（２）式で与えられるので、量子化結果ＱＦ（ｕ，ｖ）が０となる条件は以下の（１０）式で与えられる。
Ｆ（ｕ，ｖ）≧０の場合、
（Ｆ（ｕ，ｖ）−ＱＭ／２）／ＱＭ＜＋１
Ｆ（ｕ，ｖ）＜０の場合、
（Ｆ（ｕ，ｖ）＋ＱＭ／２）／ＱＭ＞−１
・・・（１０）
従って、Ｆ（ｕ，ｖ）の範囲は、以下の（１１）式で与えられる。
−ＱＭ−ＱＭ／２＜Ｆ（ｕ，ｖ）＜ＱＭ＋ＱＭ／２ ・・・（１１）
すなわち、Ｆ（ｕ，ｖ）の範囲は、以下の（１２）式を満たすことが判る。
|Ｆ（ｕ，ｖ）|＜３／２・ＱＭ ・・・（１２）
また、（１２）式と（９）式とを合わせると、以下の（１３）式が得られる。

On the other hand, since the quantization result QF (u, v) is given by the equation (2), the condition that the quantization result QF (u, v) becomes 0 is given by the following equation (10).
If F (u, v) ≧ 0,
(F (u, v) -QM / 2) / QM <+1
If F (u, v) <0,
(F (u, v) + QM / 2) / QM> −1
... (10)
Therefore, the range of F (u, v) is given by the following equation (11).
−QM−QM / 2 <F (u, v) <QM + QM / 2 (11)
That is, it can be seen that the range of F (u, v) satisfies the following expression (12).
| F (u, v) | <3/2 · QM (12)
Further, when the formula (12) and the formula (9) are combined, the following formula (13) is obtained.

ここで、画素データの関数ｆ（ｘ，ｙ）と量子化パラメータＱＭとの関係を見ると、以下の（１４）式が得られ、画素値の総和ｐは（４）式を満たすので、以下の（１５）式が得られる。

Here, looking at the relationship between the pixel data function f (x, y) and the quantization parameter QM, the following equation (14) is obtained, and the sum p of pixel values satisfies the equation (4). Equation (15) is obtained.

ｐ＜６ＱＭ ・・・（１５）
この（１５）式では、あるブロックにおける画素値の総和ｐが（１５）式を満たす場合、任意のｘ，ｙに対して離散コサイン変換および量子化後のすべてのＤＣＴ係数は零になる。

p <6QM (15)
In the equation (15), when the total sum p of pixel values in a certain block satisfies the equation (15), all DCT coefficients after the discrete cosine transform and quantization for any x and y are zero.

  Here, when each block is composed of 8 × 8 = 64 pixels, the average value of the pixel values of each block is given by p / 64. Then, considering the difference absolute value sum SS between the pixel values constituting each block and the average value of the pixel values of each block, the equation (15) can be replaced with the equation (5), and the DC coefficient of the DCT coefficient It is mathematically guaranteed that all AC coefficients except for are zero.

  As a result, it is possible to determine whether the quantization result after the discrete cosine transform becomes zero before performing the discrete cosine transform for each block, and without performing the discrete cosine transform, the discrete cosine transform and the quantization can be performed. Unnecessary blocks can be identified. Therefore, the number of blocks that need to be subjected to discrete cosine transform and quantization processing can be reduced, and the image encoding processing can be speeded up without degrading the image quality.

  In the above-described embodiment, a method of determining whether to skip the discrete cosine transform and the quantization process by determining the magnitude relationship between the difference absolute value sum SS in equation (5) and the quantization parameter QM. As described above, it may be determined whether or not to skip the discrete cosine transform and the quantization process by determining the magnitude relationship between the sum p of the pixel values of the equation (15) and the quantization parameter QM.

  However, when determining whether to skip the discrete cosine transform and the quantization process by determining the magnitude relationship between the sum p of the pixel values of Expression (15) and the quantization parameter QM, the number of blocks that can be skipped is the image Depending on the case, it will be very low, about several percent. On the other hand, when determining whether or not to skip the discrete cosine transform and the quantization process by determining the magnitude relationship between the difference absolute value sum SS in equation (5) and the quantization parameter QM, f (x , Y) -p / 64 has a value other than 0, so f (x, y) -p / 64 becomes smaller, and the number of blocks that can skip discrete cosine transform and quantization processing is increased. Can do.

  In the case of the JPEG method, the quantization parameter QM is expressed by a quantization matrix composed of 64 elements. Therefore, in order to ensure that all of the AC coefficients except for the DC coefficient of the DCT coefficient are zero, It is necessary to use the minimum value of the quantization parameter QM. On the other hand, since MPEG-4 has only one quantization parameter QM, the quantization parameter QM may be used as it is.

Alternatively, 6 of 6QM in equation (15) may be set large so that discrete cosine transform and quantization processing can be skipped even for some blocks in which AC components are not zero. As a result, the processing speed can be greatly improved without significantly affecting the image quality.
FIG. 2 is a block diagram showing a hardware configuration of an image encoding device according to an embodiment of the present invention.

  In FIG. 2, the image encoding apparatus includes a processor 11, a RAM 13, a DMA controller 14, and a pixel sum calculation unit 15, and a RAM 12 is incorporated in the processor 11. The processor 11, the RAM 13, the DMA controller 14, and the pixel sum calculation unit 15 are connected to each other via a bus 16. Here, the pixel sum calculation unit 15 can calculate the sum of the pixel values of each block before the discrete cosine transform.

  The pixel data of the frame to be encoded is stored in the RAM 13. When encoding the image data, the processor 11 instructs the DMA controller 14 to perform DMA transfer of the image data stored in the RAM 13 to the RAM 12, and at the time of DMA transfer, each processor before discrete cosine transform is performed. The sum of the pixel values of the block is calculated by the pixel sum calculation unit 15.

Then, when the image data stored in the RAM 13 is DMA-transferred to the RAM 12, the processor 11 uses the sum of the pixel values calculated by the pixel sum calculation unit 15, before performing the discrete cosine transform. The sum of absolute differences between the average pixel value of each block and each pixel value is calculated.
Then, when the processor 11 calculates the sum of absolute differences for a certain block, the processor 11 compares the sum of absolute differences for the block with a function including the quantization parameter as a variable. If the sum of absolute differences for the block is larger than the function including the quantization parameter as a variable, discrete cosine transform and quantization are performed for the block while accessing the image data stored in the RAM 12.

On the other hand, when the sum of absolute differences for the block is smaller than the function including the quantization parameter as a variable, the processor 11 skips the discrete cosine transform and the quantization for the block, and performs the discrete cosine transform and the block for the block. Without quantization, zero is output as the discrete cosine transform and quantization result for the block.
As a result, the sum of the pixel values of the block can be calculated by the pixel sum calculation unit 15 during the DMA transfer. For this reason, it becomes possible to calculate the sum of absolute differences while suppressing the load on the processor 11, and before performing the discrete cosine transform for each block, all of the AC coefficients except for the DC coefficient of the DCT coefficient are zero. Can be determined. As a result, the number of blocks that need to be subjected to discrete cosine transform and quantization processing can be reduced, and the image encoding processing can be speeded up without degrading the image quality.

The block diagram which shows the function structure of the image coding apparatus which concerns on one Embodiment of this invention. The block diagram which shows the hardware constitutions of the image coding apparatus which concerns on one Embodiment of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Skip process part, 2 Discrete cosine transform process part, 3 Quantization part, 4 Entropy encoding part, 5 Difference absolute value sum calculation part, 11 Processor, 12, 13 RAM, 14 DMA controller, 15 Pixel sum total calculation part

Claims (11)

A discrete cosine transform processing unit that performs discrete cosine transform for each block including a plurality of pixels obtained by dividing an image;
A quantization unit that performs quantization on the discrete cosine transform result obtained by the discrete cosine transform processing unit;
Based on a comparison result between an average value of pixel values of each block before the discrete cosine transform and a function including a quantization parameter used for the quantization as a variable, the discrete cosine transform and quantization for the block An image coding apparatus comprising: a skip processing unit that performs the skip processing of   The skip processing unit, based on a comparison result between an average value of pixel values of each block before the discrete cosine transform and a function including a quantization parameter used for the quantization as a variable, The image coding apparatus according to claim 1, wherein zero is output as a result of discrete cosine transform and quantization for the block without performing discrete cosine transform and quantization. A discrete cosine transform processing unit that performs discrete cosine transform for each block including a plurality of pixels obtained by dividing an image;
A quantization unit for performing quantization on the discrete cosine transform result obtained by the discrete cosine transform processing unit;
Based on a comparison result of a sum of absolute differences between the average value of the pixel values of each block before the discrete cosine transform and each pixel value and a function including a quantization parameter used for the quantization as a variable An image encoding apparatus comprising: a skip processing unit that performs a discrete cosine transform and quantization skip process on the block.   The skip processing unit is a function including, as variables, a sum of absolute differences between the average value of each pixel value of each block before the discrete cosine transform and each pixel value, and a quantization parameter used for the quantization. 4. The image according to claim 3, wherein zero is output as a result of discrete cosine transform and quantization for the block without performing discrete cosine transform and quantization for the block based on a comparison result with Encoding device. A discrete cosine transform processing unit and a processor that performs quantization processing for each block including a plurality of pixels obtained by dividing an image;
A first memory for storing image data of the image;
A second memory incorporated in the processor;
A DMA controller that DMA-transfers image data stored in the first memory to the second memory;
A pixel sum calculation unit that calculates a sum of pixel values of each block before the discrete cosine transform at the time of the DMA transfer;
The processor, based on the sum of the pixel values calculated by the pixel sum calculation unit, the sum of absolute differences between the average value of the pixel values of each block before the discrete cosine transform and each pixel value And performing discrete cosine transform and quantization skip processing on the block based on a comparison result between the calculated sum of absolute differences and a function including the quantization parameter used for the quantization as a variable. An image encoding device characterized by the above. Comparing the sum of absolute differences between the average pixel value of each block before each discrete cosine transform and each pixel value and a function including a quantization parameter as a variable;
If the sum of absolute differences for the block is greater than a function that includes the quantization parameter as a variable, performing discrete cosine transform and quantization for the block;
And a step of skipping discrete cosine transform and quantization for the block when the sum of absolute differences for the block is smaller than a function including the quantization parameter as a variable.   When the difference absolute value sum for the block is smaller than the function including the quantization parameter as a variable, the discrete cosine transform and the quantization result for the block are performed without performing the discrete cosine transform and the quantization for the block. 7. The image encoding method according to claim 6, wherein zero is output. DMA transfer of image data stored in the first memory to a second memory incorporated in the processor;
Causing the hardware to calculate the sum of the pixel values of each block before discrete cosine transform during the DMA transfer;
The sum of the pixel values calculated by the hardware is used on the processor to calculate the sum of absolute differences between the average value of the pixel values of each block before the discrete cosine transform and each pixel value. And steps to
Comparing the calculated sum of absolute differences for the block and a function including a quantization parameter as a variable on the processor;
When the sum of absolute differences for the block is larger than a function including the quantization parameter as a variable, the discrete cosine transform and quantization for the block are performed while accessing the image data stored in the second memory. Steps on the processor;
And a step of skipping discrete cosine transform and quantization for the block when the sum of absolute differences for the block is smaller than a function including the quantization parameter as a variable. Comparing the sum of absolute differences between the average pixel value of each block before each discrete cosine transform and each pixel value and a function including a quantization parameter as a variable;
If the sum of absolute differences for the block is greater than a function that includes the quantization parameter as a variable, performing discrete cosine transform and quantization for the block;
When the sum of absolute differences for the block is smaller than a function including the quantization parameter as a variable, the computer executes a step of skipping discrete cosine transform and quantization for the block Program.   When the difference absolute value sum for the block is smaller than the function including the quantization parameter as a variable, the discrete cosine transform and the quantization result for the block are performed without performing the discrete cosine transform and the quantization for the block. 10. The image encoding program according to claim 9, wherein zero is output. DMA transfer of image data stored in the first memory to a second memory incorporated in the processor;
Causing the hardware to calculate the sum of the pixel values of each block before discrete cosine transform during the DMA transfer;
Calculating a difference absolute value sum between an average value of pixel values of each block and each pixel value before the discrete cosine transform by using a sum of pixel values calculated by the hardware;
Comparing the calculated sum of absolute differences for the block with a function including a quantization parameter as a variable;
If the sum of absolute differences for the block is larger than the function including the quantization parameter as a variable, discrete cosine transform and quantization are performed for the block while accessing the image data stored in the second memory. Steps,
When the sum of absolute differences for the block is smaller than a function including the quantization parameter as a variable, the computer executes a step of skipping discrete cosine transform and quantization for the block Program.

Images