JP2002152747A - Image coder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image coder applying prediction coding to an input image that adaptively controls a coding parameter in the unit of pictures so as to attain high efficiency coding. SOLUTION: The image coder applying prediction coding to an input image is provided with: an acquisition means that acquires at least any statistic information among first statistic information that denotes the property of an input image read from a frame memory 1 such as an input image information statistic acquisition device 14, second statistic information based on correlation information in the process of prediction coding with a motion information statistics acquisition device 15 or the like and third statistic information as a result of coding or in the process of coding with a coding information statistic acquisition device 16 or the like; and a control means such as a coding control section 12 that uses a scene discrimination device 17 or the like to discriminate a scene on the basis of the statistic information acquired by the acquisition means, thereby applying adaptive control to a coding parameter in a coder 6 in the unit of frames or fields.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding apparatus for performing high-efficiency coding by switching coding parameters on a picture basis based on the characteristics of an input image and statistical information in the coding process.

[0002]

2. Description of the Related Art As one of moving picture coding methods, MP
EG-2 (Moving Picture Experts Group-2)
Has been internationally standardized, and DVD (Digital Video)
Disc / Digital Versatile Disc is applied to fields such as video contents and digital broadcasting. This M
Before PEG-2 was internationally standardized, CD-ROM (V
for recording media such as video CD) and 1.5 Mbps
MPEG-1 has been standardized as an encoding method targeting the use of lines up to the extent.

[0003] The above-mentioned MPEG-2 is an earlier MPE.
A higher image quality countermeasure than that of G-1 is incorporated. As a typical example, as a measure for encoding a higher-resolution image or interlaced image encoding, encoding by field-aware motion prediction can be performed. Encoding is also becoming more general. These functions are multiplexed into a stream (data sequence) as an additional function (extension). Basically, MPEG-2 uses M
One of the methods that has the upward compatibility with PEG-1 and can be specified in each additional function layer of MPEG-2 is MPEG-
Same as 1.

One of the additional function layers in the MPEG-2, picture, coding extens
FIG. 9 shows ion. For example, to explain one example, f code: range of motion vector that can be expressed, intra dc precision: precision of DCT DC component of the intra-frame coded block, picture structure: coding structure, top field first: input order of fields in one frame, frame pred frame dct: frame prediction frame DCT restriction flag, q scale type: quantization scale type, intra vlc format: DCT coefficient variable length code table selection, alternate scan: input order of DCT coefficients to the variable length coding unit, progressive frame: selection of whether the input signal is progressive or not.

[0005]

The parameters that can be selected in the additional function (extension) layer added to the above-mentioned MPEG-2 are basically MPEG-1.
The purpose of the present invention is to improve the efficiency in moving picture coding, which was not good at the above. However, depending on the characteristics of the input image and the coding rate, there is a case where efficient coding is possible by adaptively using the coding parameters in MPEG-1. In a conventional image coding apparatus, such an additional function layer parameter is set to a fixed value set in advance to perform predictive coding of an input image. SUMMARY OF THE INVENTION It is an object of the present invention to provide an encoding unit that enables an encoding parameter to be changed in accordance with the characteristics of an input image on a picture-by-picture basis, and enables a high-quality reproduced image to be obtained even at the same encoding rate.

[0006]

An image encoding apparatus according to the present invention is an image encoding apparatus having encoding means for performing predictive encoding of an input image, wherein the first statistical information indicating a property of the input image is provided. At least one of second statistical information based on correlation information in the process of predictive coding, and a coding result based on coding parameters or third statistical information in the coding process. Acquisition means for acquiring one piece of statistical information, for example, an acquisition means including an input image information statistical acquirer 14, a motion information statistical acquirer 15, and an encoded information statistical acquirer 16 in FIG. Encoding control means for performing a scene determination based on the statistical information obtained by the means and adaptively controlling the encoding parameters in the encoding means such as the encoder 6 in FIG. 1 in units of frames or fields, for example, Mark in Figure 1 And a reduction control section 12.

[0007] A scene determination result in which the activity of the input image is obtained as the first statistical information, the coded average quantized value is obtained as the third statistical information, and the activity and the coded average quantized value are compared. Based on this, the variable length code table in the encoding means can be switched to perform variable length encoding. Also, an average value of quantized effective coefficients per macroblock in a picture is acquired as third statistical information, and a code is determined based on a scene determination result of comparing the average value of the effective coefficients with a preset coefficient. The variable length encoding table can be switched to perform variable length encoding. Further, the average value of the horizontal component, the variance of the horizontal component, the average value of the vertical component, and the variance of the vertical component of the motion vector are acquired as the second statistical information, and the second statistical information is compared with a preset coefficient. The input scan order for variable length encoding in the encoding means can be switched based on the scene determination result. Further, an average quantization value is obtained as third statistical information, and the quantization table is switched by the encoding means based on a scene determination result of comparing the average quantization value with a preset coefficient to perform quantization. Can be done.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is an explanatory diagram of an embodiment of the present invention. 1 is a frame memory, 2 is an original MB (macroblock) reading section, 3 is a reference block reading section, and 4 is a motion vector search. , 5 is a prediction decision unit, 6 is an encoder, 7 is a local decoder, 8 and 9 are switching units, 10 is an adder, 11 is a subtractor, 12 is an encoding control unit, and 13 is header information generation. Reference numeral 14 denotes an input image information statistic acquisition unit, 15 denotes a motion information statistic acquisition unit, 16 denotes an encoding information statistic acquisition unit, and 17 denotes a scene determination unit.

A frame memory 1 includes an area for storing input image information and an area for storing reference image information.
, And read by the reference block reading unit 3 in macroblock units within the search range of the reference image area, and a motion vector is obtained by the motion vector search unit 4 and input to the prediction determination unit 5. Also, the encoder 6 has a DCT (Discret
e Cosine Transform), quantization,
It includes a function with variable-length coding. In addition, MPE
The macroblock (MB) size in the G-2 system is 1
The block to be subjected to DCT is 6 × 16 pixels, and the macroblock is divided into four to be 8 × 8 pixels.

The switching units 8 and 9 switch to the adder 10 and the subtractor 11 when encoding within a frame and between frames, and adder 10 and a subtractor when encoding between a field and between fields. It switches to the side opposite to the eleventh side.
The local decoder 7 includes functions of inverse quantization and inverse DCT, performs a decoding process using the quantized output of the encoder 6 before the variable encoding, and reconstructs the reference image. And stores it in the area of the reference image information in the frame memory 1.

The DCT in the encoder 6 is a two-dimensional DC
T, as described above, in MPEG-2, 8
This is performed for a block of × 8 pixels. This block
If (x, y) and the coefficient of the DCT result are F (u, v), the following equation (1) is obtained.

(Equation 1)

By this DCT operation, the image information in units of blocks is converted into frequency components, and the effective components are gathered on the low frequency component side, thereby reducing the coded information. [X], the quantization scale is Qs, the quantization matrix value is Qm [x],
The quantization result is Level [x], and the value of the process is Level
el ′ [x], a DC component (Intra DC) of a quantization result in the intra coding and an AC component (I
ntra AC) and a calculation process for obtaining a quantization result (Non Intra) in non-intra encoding are as follows.
It can be expressed by the following equation (2). In addition, p = 3 and q = 4 are generally used for p and q in the formula. "//" indicates an operation for rounding a fraction of a division result.
“/” Indicates an operation for rounding down a fraction of the division result.

(Equation 2)

In order to perform inter-frame coding, it is necessary to obtain inter-frame difference information. For this reason, the coded data is decoded by a local decoder 7 by inverse quantization and inverse DCT processing. Then, the image is reconstructed and stored in the frame memory 1 as a reference image. As described above, the local decoder 7 includes the processing functions of inverse quantization and inverse DCT. The inverse quantization can be performed by the processing shown in equation (3). This can be performed by the processing shown in equation (4).

(Equation 3)

When actually performing an input image encoding process,
In the first picture (frame or field), since there is no picture to be referenced, intra-picture coding is performed, and inter-picture coding is performed from the next picture. Note that intra-picture encoding is generally performed at predetermined intervals for the purpose of periodic refresh. In the inter-picture coding, motion prediction is performed by the motion vector searcher 4. For example, FIG.
As shown in the figure, for each pixel between the macroblock 22 of the original image 21 and the macroblock in the search range 24 of the reference image 23, a position where the accumulated value of the absolute difference is minimum is searched for a motion vector. , Multiplexed with encoded information.

An encoding means is constituted by the functional parts 1 to 11 in FIG. 1 described above, and an encoding control section 12 for controlling an encoding parameter in the encoder 6 controls the encoding control means. And the header information including the encoding parameter is generated by the header information generation unit 13, and the encoder 6
The header data is added to the coded data in picture units from, and transmitted.

The input image information statistical acquisition unit 14 is means for acquiring first statistical information indicating the characteristics of the input image.
For example, statistical information (activity) on a luminance signal is obtained as characteristic information of the current picture stored in the frame memory 1. In this case, the luminance value of each pixel in the input frame is accumulated and divided by the accumulated number of pixels, so that an average frame luminance can be obtained and used as the first statistical information. That is, the pixel set in the frame is U, and the luminance is Pixel. i, number of pixels is Num i, average frame luminance AveY, frame luminance variance VarY
Then, it is expressed by the following equations (5) and (6). Note that A in the equation (6) indicates the frame luminance average AveY.

(Equation 4)

The motion information statistic acquiring unit 15 is means for acquiring second statistical information indicating correlation information between frames or between fields. For example, the motion information statistic acquiring unit 15 calculates the motion information obtained by the macroblock in the motion vector searching unit 4. By accumulating the vectors and dividing by the number of macroblocks, an average value of the motion vectors can be obtained. Alternatively, the variance of the motion vector can be obtained by calculating the sum of squares of the difference between the motion vector and the average value and dividing the sum by the number of macroblocks.

That is, the set of macroblocks in a frame is V, and the horizontal and vertical components of each motion vector are Vec.
H i, VecV i, and the horizontal component average is AveH
V, the horizontal component variance is VerHV, and the vertical component average is Av
If eVV and the vertical component variance are VerVV, (7) to
It is expressed by equation (10).

(Equation 5) Note that AH in the equation (8) is a horizontal component average AveHV.
And AV in equation (10) is the vertical component average Av
Indicates eVV.

The encoding information statistical acquisition unit 16 is means for acquiring third statistical information in the encoding process. For example, the encoding information statistical acquiring unit 16 accumulates information as a result of encoding each macroblock, And the average value of the quantization values. In that case, the set of macroblocks in the frame is V, and the amount of generated information of each macroblock is Bit. i, SumB is the amount of generated information of the picture, and Qs is the quantization scale value of each macroblock. i, the average quantization value is AveQ, the amount of information generated Sum of the picture and the average quantization value AveQ
Is represented by Expressions (11) and (12).

(Equation 6)

The effective coefficient of each macroblock after quantization is represented by Coef i, assuming that the average value of effective coefficients per macroblock in a picture is AveC, the average effective coefficient AveC is expressed by equation (13).

(Equation 7)

The scene determination unit 17 performs a scene determination based on at least one of the first, second, and third statistical information described above. As a means for acquiring the first statistical information, , The frame luminance average AveY and the variance VarY from the input image information statistical acquisition unit 14 and the horizontal component average AveHV, the horizontal component variance VerHV, and the vertical component from the motion information statistical acquisition unit 15 as means for acquiring the second statistical information. Average AveVV, vertical component variance AerVV, and third
Using one or more of the picture generation information amount SumB, the average quantization value AveQ, the effective coefficient average value AveC, etc. from the coding information statistics acquisition unit 16 as a means for acquiring the statistical information of A decision is made on a scene, a scene with a flat luminance, or the like, and the encoding control unit 12 adaptively controls the encoding parameters so that the encoder 6 encodes the input image.

FIG. 3 is a flowchart of the determination process according to the embodiment of the present invention. An input image is read from the frame memory 1 (a1), and a header is generated for each picture by the header information generator 13 (a2). Then, a motion search is performed in the motion vector searcher 4 (a3), and MB (macroblock) coding is performed in the encoder 6 (a4).
It is determined whether or not the picture End, that is, one picture has been completed (a5). If not completed, the process proceeds to step (a3), and if completed, necessary information is obtained (a10).

The first statistic information is obtained by the input image information statistic acquisition unit 14 (a6), and the motion information statistic acquisition unit 15 obtains the first statistic information based on the motion vector search result as the second statistic information. The obtained statistical information is obtained (a7), and the result of encoding by the encoder 6 or the third statistical information in the encoding process is acquired by the encoded information statistical acquirer 16 (a8). When each average value is obtained, an averaging process is performed (a9). The result of this averaging process is determined by the scene determiner 1
In 7, it is acquired as necessary information acquisition (a10), and it is determined whether or not a predetermined condition is satisfied (a11). Based on the determination result, selection of encoding parameter 1 (a12) or encoding parameter 2 (a13) is performed. A selection is made, and it is determined whether or not the coding is End (a14). If not completed, the process proceeds to step (a1). Note that the determination step (a11)
In this case, it is also possible to control selection switching of a larger number of encoding parameters in accordance with a plurality of types of determination conditions.

For example, an additional function (pictu) shown in FIG.
re coding extension) as intra vlc format (DCT
When performing adaptive control (selection of a variable length code table for coefficients), intra vlc format = 0 and int
ra vlc By setting format = 1, ta in FIG.
ble = 0 and table = 1 in FIG. 5 can be selected. 4 and 5 show variable length codes (Va
removable length code and run (Ru)
2 shows a part of a variable length code table including n) and a level (level), and s of the last bit is a sign of the level, and 0 is positive and 1 is negative. Also, 1s indicates the first DCT coefficient of the block, and 11s indicates the next DCT coefficient.

Also, as compared with table = 0 in FIG. 4, table = 1 in FIG. 5 makes it possible to assign a variable length code by evenly shortening a bit to a certain extent. When there are many coefficients, FIG.
It is more effective to select table = 1 and perform variable length coding. Also, in practice, the number of effective coefficients decreases as the average quantization value increases, and the number of effective coefficients decreases when the activity is small on a flat screen. Even when the average quantization value is large, the number of effective coefficients is small when the activity is small, and the number of effective coefficients is large when the activity is large.

FIG. 6 is a diagram for explaining the selection of a variable length code table. vlc When format = 0,
For intra blocks and non-intra blocks, table = 0, intra vlc format =
When 1, the table =
For 1, non-intra blocks, table = 0 is selected, and variable length coding is performed.

AveQ> VarY * α ₁ + β ₁ (14) using the average quantization value AveQ obtained by the above equation (12) and VarY as the activity of the input image obtained by the equation (6). Is satisfied, the number of effective coefficients is small.
ble = 0 is selected, and if not established, table =
Switching control is performed adaptively so as to select 1. In addition,
α ₁ and β ₁ indicate weighting coefficients.

Further, for further simplification, it is determined whether or not the condition of AveC <α ₂ (15) is satisfied by using the average value AveC of the effective coefficients obtained by the equation (13). If tab
If le = 0 is selected, if not established, table = 1
Is adaptively controlled so as to select, and variable-length coding can be performed. Incidentally, alpha ₂ represents a coefficient. That is,
The fact that the average value AveC of the effective coefficients per macroblock is smaller than the preset coefficient α ₂ corresponds to a case where the activity of the input image is small, and performs variable length coding by switching to select table = 0. The encoding efficiency is improved.

[0029] Altenate scan (input order of DCT coefficient variable-length coding)
itenate When scan = 0, the order of scanning the DCT coefficients is zigzag scan (scan type 0) shown in FIG. s
When can = 1, an alternate scan (scan type 1) shown in FIG. 7B can be performed. In this case, as compared with (A) scan type 0, (B)
Scan type 1 preferentially encodes the coefficient of the vertical component in the frequency component. Factors that cause such a large number of effective coefficients in the vertical component include, for example, a case where the odd / even field image in the encoding of the interlaced image is significantly different, that is, a case where there is movement such as panning or tilt. .

MPEG-2 is capable of performing field DCT, which cuts out every other line when cutting out a block from a macroblock, as an input to DCT, for predicting motion between fields in consideration of fields.
Also, as shown in FIG. 2, the motion vector search range for performing motion prediction is generally a predetermined range 24 centered on the 0 vector. Therefore, when performing field prediction, the search range is insufficient. In the case of such a large movement, the distribution of the effective coefficient has more frequency vertical component directions.

Therefore, the horizontal component average value Ave of the equation (7)
Using the HV and the horizontal component variance VerHV of the equation (8), it is determined whether or not the condition of AveHV> α ₃ (16) VerHV <β ₃ (17) is satisfied. In the case of a large movement in the horizontal direction to some extent, which satisfies this condition, the vertical component is increased.
scan = 1, otherwise, altatenate
Adaptive length control is performed so as to select scan = 0, and variable length coding is performed. Note that α ₃ and β ₃ indicate constants, and the motion search range is generally different depending on the device.
It can be set in advance based on the motion search range.

As the third statistical information obtained by the coding information statistical obtaining unit 16, a quantization scale type q using an average quantization value AveQ is used. scale A selection of type can be made. That is, for the quantization table shown in FIG. scale When type = 0, the change in the quantization scale value is linear, and the quantization scale value (puntiser) is changed. scale code) is 2
~ 62. On the other hand, q scale
In the case of type = 1, a non-linear change is performed to cover a wide range of quantization scale values. Fine portions of quantization are finer and coarse portions of quantization are more coarsely quantized. To change.

If the quantization scale value becomes extremely small or
If is not too large, the quantization table
There is no significant difference when choosing either
When the average quantization value is large when
Has q scale select type = 1
Control. For example, the average quantization value Av according to the equation (12)
For eQ, AveQ <α_Four … (18) AveQ> β_Four It is determined whether the condition of (19) is satisfied, and this condition is satisfied.
In other words, when the average quantization value is extremely small,
If the edge is large, the quantization scale value changes nonlinearly.
Q scale Select type = 1, and
If outside, q scale I will select type = 0
Adaptive control.

The present invention is not limited to the above-described embodiments, but can be variously added and changed. The coding parameter by switching the quantization table or the variable length code table is described. In addition to the adaptive switching control of (i), it is also possible to perform switching control of other encoding parameters, and obtain statistical information of a picture to be encoded before encoding the picture, and perform feedforward control. Encoding can also be performed. When the method is applied to storage media, provisional encoding of a picture is performed using a plurality of types of encoding parameters, a final decision is made based on the provisional encoding result, and header information including the optimal encoding parameters is encoded. It can also be added to data and stored.

[0035]

As described above, according to the present invention, the first statistical information such as the activity indicating the property of the input image,
Second statistical information such as a motion vector of correlation information in the process of predictive coding, and third statistical information such as a coding information amount of a coding result based on a coding parameter and a quantized average value of the coding process. Using at least one or a plurality of pieces of statistical information, the coding parameters in the coding means are controlled in units of frames or fields, so that the input information can be input without increasing the amount of coding information. There is an advantage that an image can be optimally encoded.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram of an embodiment of the present invention.

FIG. 2 is an explanatory diagram of a motion search.

FIG. 3 is a flowchart of a determination process according to the embodiment of the present invention.

FIG. 4 is an explanatory diagram of a variable length code table.

FIG. 5 is an explanatory diagram of a variable length code table.

FIG. 6 is an explanatory diagram of selection of a variable length code table.

FIG. 7 is an explanatory diagram of an input scan.

FIG. 8 is an explanatory diagram of a quantization table.

FIG. 9 is an explanatory diagram of an example of an additional function.

[Explanation of symbols]

 Reference Signs List 1 frame memory 2 original MB reading unit 3 reference block reading unit 4 motion vector searcher 5 prediction decision unit 6 encoder 7 local decoder 8, 9 switching unit 10 adder 11 subtractor 12 encoding control unit 13 header information Generating unit 14 Input image information statistics acquisition unit 15 Motion information statistics acquisition unit 16 Encoding information statistics acquisition unit 17 Scene determination unit

 ────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Kiyoshi Sakai 4-1-1, Kamidadanaka, Nakahara-ku, Kawasaki-shi, Kanagawa F-term within Fujitsu Limited (reference) 5C059 MA00 MA04 MA05 MA23 MC24 MC32 MC34 ME01 NN03 NN28 RB09 RB14 SS02 SS06 TA47 TA58 TB05 TC06 TC10 TC12 TD03 TD04 UA02 UA33

Claims (5)

[Claims] 1. An image encoding apparatus having encoding means for performing predictive encoding of an input image, comprising: a first statistical information indicating a characteristic of the input image; and a correlation in a process of the predictive encoding. Acquisition means for acquiring at least one of the second statistical information based on the information and at least one of the encoding result based on the encoding parameter or the third statistical information in the encoding process; An image control unit that performs a scene determination based on the statistical information acquired by the acquisition unit and adaptively controls an encoding parameter in the encoding unit on a frame or field basis. Encoding device. 2. The method according to claim 1, wherein the obtaining unit obtains an activity of the input image as the first statistical information, and obtains an encoded average quantized value as the third statistical information. The encoding control means has a control structure for performing variable length encoding by switching a variable length code table in the encoding means based on a scene determination result comparing the activity with the encoded average quantized value. The image encoding device according to claim 1, wherein: 3. The method according to claim 2, wherein the obtaining unit obtains an average value of quantized effective coefficients per macroblock in a picture as the third statistical information, and the encoding control unit performs It has a control structure for performing variable length coding by switching a variable length code table in the encoding means based on a scene determination result comparing an average value of effective coefficients with a preset coefficient. The image encoding device according to claim 1. 4. The apparatus according to claim 1, wherein the obtaining unit obtains, as the second statistical information, an average value of a horizontal component, a variance of a horizontal component, an average value of a vertical component, and a variance of a vertical component of the motion vector. The encoding control means has a control configuration for switching an input scan order for variable-length encoding in the encoding means based on a scene determination result comparing the second statistical information with a preset coefficient. The image encoding device according to claim 1, wherein: 5. The method according to claim 1, wherein the obtaining unit obtains an average quantization value as the third statistical information, and the encoding control unit compares the average quantization value with a preset coefficient. 2. The image encoding apparatus according to claim 1, further comprising a control structure for performing quantization by switching a quantization table in said encoding means based on a scene determination result.