JP2002209216A - Image coding device and image coding method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent the number of required bits for coding a motion vector from extremely increasing even when a predictive vector can not be calculated with sufficient accuracy. SOLUTION: In the case that each predictive vector sufficiently comes true at each motion vector and difference vectors between the predictive vectors and the motion vectors are distributed with a concentration to around zero in a VLC(Variable-Length Coding) unit 16, an F-code is decided from a maximum value of the motion vectors so that the ratio of a fixed length coding part is made small. In the case that the prediction accuracy of the predictive vector is insufficient and difference vectors between the predictive vectors and the motion vectors are not distributed with the concentration to around zero, the F-code is decided larger than a minimum required value so that the ratio of a variable length coding part is made small.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to an image encoding apparatus and an image encoding method, and more particularly to an image encoding apparatus and an image encoding method suitable for transmitting and receiving moving image data via a transmission path.

[0002]

2. Description of the Related Art For example, in a system for transmitting moving image data to a remote place, such as a video conference system or a video telephone system, the moving image data is line-correlated or inter-frame-connected in order to use a transmission path efficiently. Data is transmitted and received with a reduced data capacity by performing compression encoding using correlation.

As an example of a high-efficiency coding method for a moving image, M
There is an MPEG standardized by the Moving Picture Experts Group (PEG). In the MPEG system, motion-compensated prediction coding and discrete cosine (Discrete Cosine Tran
sform; DCT) coding. In the MPEG system, several profiles and levels are defined to support various applications and functions. Among them, the most basic is the main profile main level (Main Pro
file at Main Level; MP @ ML).

In a conventional image coding apparatus, a prediction vector is calculated for each macroblock, and coding is performed on a difference vector obtained as a difference value between the prediction vector and a motion vector to be coded. . The difference vector between the motion vector and the prediction vector is represented by a flag value indicating the encoding range of the motion vector set for each frame, and a general part of a motion vector having a variable bit length (variable length encoding unit) and a fixed bit. It is divided and encoded into a detailed part (fixed-length encoding part) of a motion vector having a length.

[0005] MPEG1, MPE based on MPEG system
In G2, MPEG4, and the like, a motion vector prediction vector is calculated using the motion vector of a macroblock closer to the position of the macroblock to be encoded.
It is calculated based on the methods defined in EG1, MPEG2, and MPEG4, respectively.

[0006]

However, the VLC used when coding the generalized part of the decomposed motion vector is used.
The (Variable-Length Coding) table is created on the assumption that the accuracy of the prediction vector is sufficient. That is, the VLC table is created on the assumption that the difference vector calculated as the value of the difference between the prediction vector and the motion vector is concentrated near zero.

However, in the case of an image in which the luminance change is small over the entire image and the luminance value is constant in a certain frame, or in an image such as a river flow, the correlation between motion vectors between adjacent macroblocks. Often decreases. At this time, the difference vector between the prediction vector calculated from the motion vector between neighboring macroblocks and the motion vector of the macroblock to be actually coded is large.

[0008] Therefore, in the conventional image coding apparatus, when VLC coding is performed on a motion vector, the code length of the coded code frequently becomes large. And the image quality of the encoded image is degraded.

The present invention has been proposed in view of such a conventional situation, and prevents an increase in the amount of generated motion vector bits even when the prediction vector cannot be calculated with sufficient accuracy. It is an object of the present invention to provide an image encoding device and an image encoding method that enable the above.

[0010]

An image coding apparatus according to the present invention detects a motion vector from input image processing information, and divides the motion vector into a variable length coding section and a fixed length coding section to perform coding. In the image encoding device, a correlation value of a motion vector in the image based on the image processing information is detected, and a flag value indicating a coding range of the motion vector is calculated using the correlation value and the maximum value of the motion vector in the image. The above-mentioned object is achieved by providing the flag value determining means for determining.

Here, the flag value determining means calculates a correlation value from the motion vector and the prediction vector. If the correlation value is concentrated near zero, the flag value is determined so as to reduce the ratio of the fixed length coding unit. The flag value is determined based on the maximum value of the vector, and when the correlation value is dispersed from near zero, the flag value is determined based on the maximum value of the motion vector so that the number of variable length coding units is reduced. There are things.

An image encoding method according to the present invention comprises the steps of: detecting a motion vector from input image processing information; dividing the motion vector into a variable length encoding unit and a fixed length encoding unit; A flag value determining step of detecting a correlation value of a motion vector in the image based on the image processing information and determining a flag value representing a coding range of the motion vector using the correlation value and the maximum value of the motion vector in the image With the provision of the above, the above-mentioned object is achieved.

Here, in the flag value determining step, the correlation value is calculated from the difference between the motion vector and the prediction vector, and when the correlation value is concentrated near zero, the ratio of the fixed length coding unit is reduced. The flag value is determined based on the maximum value of the motion vector so as to decrease, and when the correlation value is dispersed from near zero, the flag value is determined based on the maximum value of the motion vector so that the number of variable-length coding units decreases. The flag value is determined.

[0014]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an example of a configuration of an image coding apparatus according to an embodiment of the present invention.

The image encoding apparatus shown as an embodiment of the present invention has the advantage that even when the correlation between motion vectors in an image (hereinafter, referred to as a frame) is low and the calculation accuracy of a prediction vector is reduced, the motion vector is not changed. A correlation value is calculated from the vector and the prediction vector, and when the correlation value is concentrated near zero, the flag value is determined based on the maximum value of the motion vector so that the ratio of the fixed-length coding unit is reduced. However, when the correlation values are scattered from near zero, the flag value is determined based on the maximum value of the motion vector so that the ratio of the variable length coding unit is reduced. This makes it possible to prevent an excessive increase in the required number of bits.

An image coding apparatus 1 shown as one configuration example of an embodiment of the present invention has an MPEG (Moving Picture Experer).
This is an image encoding device that encodes image information at a main profile at a main level (MP @ ML) based on a ts Group method. As shown in FIG. A first frame memory 11 and a second frame memory 21 for storing, a motion vector detector 12 for detecting a motion vector, and a first calculator 1 for calculating a difference between image data and predicted image data
3, a second computing unit 20, a discrete cosine transformer (hereinafter referred to as a DCT unit) 14 for performing a discrete cosine transform,
A quantizer 15 that converts DCT coefficients into quantized coefficients in accordance with a quantization step, a variable-length encoder (hereinafter, referred to as a VLC unit) 16 that performs variable-length encoding, and stores encoded data. A buffer 17 and an inverse quantizer 1 for converting a quantized coefficient into a DCT coefficient according to a quantization step
8, an IDCT unit 19 for performing an inverse discrete cosine transform, and a motion compensator 22 for extracting predicted image data.

In the image encoding device 1, a first frame memory 11 receives image data to be encoded,
Remember temporarily.

The motion vector detector 12 reads out the image data stored in the first frame memory 11 in units of macroblocks composed of, for example, 16 × 16 pixels, and detects a motion vector. Here, the motion vector detector 12 converts the image data of each frame into an intra-frame coded image (hereinafter, referred to as an I picture), a forward prediction coded image (hereinafter, referred to as a P picture), and bidirectional prediction coding. The image is processed as any one of the images (hereinafter, referred to as a B picture). Which of an I picture, a P picture, and a B picture is to be processed as an image of each sequentially input frame is determined by, for example, I, B, P, B, P,. Are determined in advance so as to be processed in the order of

That is, the motion vector detector 12 refers to a predetermined reference frame among the image data stored in the first frame memory 11 and becomes a target of encoding with the referenced frame. Frame 1
By performing block matching (pattern matching) with a macroblock (small block) of 6 pixels × 16 pixel lines, a motion vector of the macroblock is detected.

In MPEG, four types of prediction modes, i.e., intra coding (intra-frame coding), forward prediction coding, backward prediction coding, and bidirectional prediction coding, are provided as image prediction modes representing the format of image prediction. I have. An I picture represents intra-coded image data, a P picture represents image data encoded by any of intra coding and forward prediction coding, and a B picture represents intra coding and forward prediction coding. It is determined to represent image data encoded by any one of encoding, backward prediction encoding, and bidirectional prediction encoding.

The motion vector detector 12 sets intra coding as a prediction mode for the I picture.
In this case, the motion vector detector 12 does not detect a motion vector, and sets the intra prediction mode to a variable length encoder (hereinafter, referred to as a VLC unit) 16 and a motion compensator 22.
Output to

The motion vector detector 12 performs forward prediction on the P picture and detects the motion vector. Further, the motion vector detector 12 compares a prediction error caused by performing forward prediction with, for example, a variance of a coding-target macroblock (P-picture macroblock).

As a result, when the variance of the macroblock is smaller than the prediction error, the motion vector detector 12 sets the intra coding as the prediction mode and outputs the motion vector to the VLC unit 16 and the motion compensator 22. I do. If the variance of the macroblock is larger than the prediction error, the motion vector detector 12 sets forward prediction encoding as the prediction mode, and outputs the predicted motion vector to the VLC unit 16 and the motion compensator 22 together with the detected motion vector. I do.

Further, the motion vector detector 12 performs forward prediction coding, backward prediction coding,
And bidirectional predictive coding to detect respective motion vectors. The motion vector detector 12 detects a prediction error of the forward prediction, the backward prediction, and the bidirectional prediction (hereinafter, appropriately referred to as a minimum prediction error), and determines the minimum prediction error and the encoding target. Are compared with, for example, the variance of the macroblock (macroblock of the B picture).

As a result, if the variance of the macroblock is smaller than the minimum prediction error, the motion vector detector 12
Sets intra coding as a prediction mode, and outputs a motion vector to the VLC unit 16 and the motion compensator 22.
If the variance of the macroblock is larger than the minimum prediction error, the motion vector detector 12 sets the prediction mode in which the minimum prediction error is obtained as the prediction mode,
Output to the VLC unit 16 and the motion compensator 22 together with the corresponding motion vector.

The first arithmetic unit 13 stores the same macroblock as the image data read from the first frame memory 11 by the motion vector detector 12 in the first frame memory 1.
1 and its macroblock and motion compensator 22
Is calculated from the predicted image data. The value of this difference is supplied to the DCT unit 14.

The DCT unit 14 performs DCT (Discrete Cosine Transform) processing on the image data from the first computing unit 13. The DCT coefficient obtained as a result is supplied to the quantizer 15.

In the quantizer 15, a quantization step (quantization scale) is set in accordance with the amount of data stored in the buffer 17, the amount of data stored in the buffer 17, and the buffer feedback. In the step, the DCT coefficient from the DCT unit 14 is quantized. The quantized DCT coefficient (hereinafter, appropriately referred to as a quantization coefficient) is transmitted to the VLC unit 1 together with the set quantization step.
6.

In the VLC unit 16, the quantized coefficients supplied from the quantizer 15 are converted into a variable length code such as a Huffman code and output to the buffer 17. further,
The VLC unit 16 performs a quantization step from the quantizer 15,
A prediction mode from the motion vector detector 12, that is, a prediction mode and a motion vector selected from intra coding (intra-frame predictive coding), forward predictive coding, backward predictive coding, or bidirectional predictive coding. Variable-length coding is performed, and the resulting coded data is supplied to the buffer 17.

The buffer 17 temporarily stores the encoded data from the VLC unit 16, smoothes the data amount of the encoded data, and outputs the encoded data as an encoded bit stream. Alternatively, it is recorded on a recording medium (not shown). The buffer 17 supplies the data storage amount of the encoded data to the quantizer 15. The quantizer 15 includes a buffer 17
The quantization step is set according to the data storage amount of the encoded data from. That is, when the buffer 17 is about to overflow, the quantizer 15 increases the quantization step to reduce the data amount of the quantization coefficient, and when the buffer 17 is about to underflow, the quantization step decreases the quantization step to reduce the quantization coefficient. Increase the amount of data. Thus, the quantizer 15 prevents the buffer 17 from overflowing and underflowing.

The quantization coefficient and the quantization step output from the quantizer 15 are determined not only by the VLC unit 16 but also by the inverse quantizer 1.
8 as well. In the inverse quantizer 18, the quantizer 15
Are inversely quantized in accordance with the quantization step from the quantizer 15 to be converted into DCT coefficients. This DCT coefficient is calculated by using an inverse DCT unit (hereinafter, I
It is referred to as DCT unit. ) 19.

In the IDCT unit 19, the DCT coefficient is
T processing, and the resulting data
10 is supplied. For intra coding, I
The data from the DCT unit 19 passes through the second arithmetic unit 20 and is directly sent to the second frame memory 21 as locally decoded image data (hereinafter, referred to as locally decoded image data). Supplied. This locally decoded image data is the same as the image data obtained on the receiving side.

The locally decoded image data obtained by the second arithmetic unit 20 is stored in the second frame memory 21 and then subjected to forward prediction coding, backward prediction coding, and bidirectional prediction coding. Is used as reference image data.

The motion compensator 22 includes a motion vector detector 1
2, the local decoded image data stored in the second frame memory 21 is read in accordance with the prediction mode and the motion vector, and the read image data is read as the predicted image. The data is supplied to the first computing unit 13 and the second computing unit 20 as data.

The motion compensator 22 includes the motion vector detector 1
When only the prediction mode is received from No. 2, especially when the prediction mode is the intra-coding mode, no predicted image data is output. At this time, the first arithmetic unit 13 and the second arithmetic unit 20 directly output the macroblock read from the first frame memory 11 to the DCT unit 14 without performing the arithmetic processing.

Here, the configuration of the VLC unit 16 will be described in detail with reference to FIG. Specifically, the VLC unit 16 includes an MV buffer 31 that stores the input vector and the motion vector in one frame, and converts the motion vector into a general part (variable length coding part) and a detailed part (fixed length coding). A motion vector decomposer 32 for decomposing the motion vector and a motion vector general part coder 33 for encoding a general part of the decomposed motion vector.
A motion vector detail unit encoder 34 for encoding a detail portion of the decomposed motion vector; an F_code detector 35 for determining an F_code flag indicating a range in which the motion vector can be encoded; The prediction vector generator 36 calculates a value, and an adder 37 generates a difference value between the motion vector input from the MV buffer 31 and the prediction vector generated by the prediction vector generator 36. .

The MV buffer 31 is temporary storage means for storing the input motion vector and the motion vector in one frame. The motion vector accumulated in the MV buffer 31 is calculated by a prediction vector creator 36, an adder 37, and F_co.
It is output to the de detector 35.

The motion vector decomposer 32 receives a difference vector from an adder 37 described later and an F_code from an F_code detector 35 described later, and divides the input vector into a general part and a detailed part. The details of the division processing operation in the motion vector decomposer 32 will be described later.
For the input difference vector, mv is a motion vector, residual
Is the motion vector detail part, and vlc_code_magnitude is the motion vector outline part. if (mv == 0) {residual = 0; vlc_code_magnitude = 0;} else {int scale factor = 1 <<(fcode-1); residual = (abs (mv) -1)% scale_factor; vlc_code_magnitude = ((abs (diff_vector) -residual + (scale_factor-1)) / scale_factor × sign (mv);} The motion vector decomposer 32 outputs the motion vector rough part obtained by the decomposition to the motion vector rough part encoder 33, The detailed portion of the motion vector is output to the motion vector detailed portion 34.

The motion vector outline section encoder 33 encodes the input motion vector outline section based on a VLC (variable length coding) table shown in Table 1 below.
The bit stream obtained by encoding is output to the motion vector detail part encoder 34.

[0040]

[Table 1]

The motion vector detail part encoder 34 encodes the inputted motion vector detail part as fixed bits. At this time, the motion vector detail part is (F_code-1) b
It is encoded with it. The encoded bits are output as part of a bit stream. If the F_code flag is 1, the motion vector detail part encoder 34
Produces no output.

The F_code detector 35 determines the F_code flag with reference to the motion vectors of all macroblocks of one frame stored in the MV buffer 31. The F_code flag indicates the range in which the motion vector can be encoded.
Are defined as shown in FIG.

[0043]

[Table 2]

For example, if the value of F_code is 1, [-1
6,15.5].
Normally, F_code is set according to the maximum value of the motion vector in the frame, and the minimum F_code value covering the maximum motion vector is set. The F_code value determined in this way is output from the F_code detector 35 to the motion vector decomposer 32.

The prediction vector generator 36 calculates a prediction value for the motion vector of each macroblock based on the information of the motion vector stored in the MV buffer 31. The created prediction vector is output to the adder 37. When calculating the prediction vector, a method determined for each image coding method is used.

The adder 37 calculates M
The motion vector to be coded is input from the V buffer 31, and the predicted value of the motion vector of the corresponding macro block is input from the predicted vector creator 36. The adder 37 creates a difference value (hereinafter, referred to as a difference vector) between the motion vector input from the MV buffer 31 and the prediction vector created by the prediction vector creation unit 36, and decomposes the difference vector into a motion vector. Output to the unit 32. As described above, the image encoding device 1 compresses and encodes the moving image data using the line correlation, the inter-frame correlation, and the like.

Subsequently, the VLC having the above configuration
The operation of the unit 16 for decomposing the input difference vector into a general part and a detailed part will be described in detail.

The motion vector input to the VLC unit 16 is input to the MV buffer 31. MV buffer 31
Then, the input motion vector is stored and sent to the predicted vector creator 36 and the adder 37 as needed.

The prediction vector creator 36 receives the motion vector from the MV buffer 31 and
A prediction vector of each macro block is created using the motion vector information stored in the MV buffer 31 based on each standard of the image coding scheme of EG2 or MPEG4. The prediction vector creator 36 outputs the prediction vector calculated for each macroblock to the F_code detector 35 and the adder 3.
7

The F_code detector 35 detects the maximum value of the input motion vector, and based on the maximum value of the detected motion vector, the F_code detector 35 includes all the motion vectors.
Detect code. Further, the F_code detector 35 compares the motion vector with the motion vector output from the motion vector generator 36 and detects whether there is a sufficient correlation between the motion vector and the motion vector in the frame.

When the F_code detector 35 determines that the correlation between the motion vector and the prediction vector is sufficiently high over the entire frame, the F_code detector 35 sends the F_code determined based on the maximum value of the motion vector to the motion vector decomposer 32. Is output.

When the F_code detector 35 determines that the correlation between the motion vector and the prediction vector is low over the entire frame, the F_code detector 35 has a value larger than the F_code determined based on the maximum value of the motion vector described above. Change to F_code, and output the changed F_code to motion vector decomposer 32.

That is, in the image encoding device 1, VLC
When the prediction vector sufficiently satisfies each motion vector and the vector after the difference is concentrated near zero, the F_code is larger than the maximum value of the motion vector so that the ratio of the fixed-length coding unit is reduced. To determine. Further, when the prediction accuracy of the prediction vector is insufficient and the vector after the difference is not concentrated near zero, the value of the F_code is required at a minimum so that the ratio of the variable length coding unit is reduced. Encoding is performed by setting a value larger than the value. As described above, the image coding apparatus 1 is characterized in that the value of F_code is changed based on the correlation between the detected prediction vector and the motion vector.

The F_code output by the above means is output to the motion vector decomposer 32 and the motion vector detail unit encoder 34.

The adder 37 creates a difference vector from the motion vector output from the MV buffer and the prediction vector output from the prediction vector generator 36.
This difference vector is sent to the motion vector decomposer 32.

The motion vector decomposer 32 decomposes the difference vector from the adder 37 into a general part showing the outline of the differential vector and a part showing the detailed part by the operation shown in the conventional example, and Is output to the motion vector outline part encoder 33, and the detailed part of the difference vector is output to the motion vector outline part encoder 34.

The motion vector outline part encoder 33 receives the outline part of the difference vector output from the motion vector decomposer 32 and encodes it using a VLC table. The outline of the encoded motion vector is output as a bit stream.

On the other hand, the motion vector detail part encoder 34
Is the detailed part of the difference vector output from the motion vector decomposer, and the F_code output from the F_code detector 35.
Is input, and the detailed part of the difference vector is encoded with (f_code-1) fixed bits, and the resultant is output as a bit stream.

As described above, the VLC unit 16 calculates F_co based on the correlation between each motion vector and the prediction vector.
It is possible to change the value of de.

When the value of F_code is changed, F_code may be increased by a fixed value based on the maximum value of the motion vector, or may be changed according to the degree of correlation between the motion vector and the prediction vector. By making the increment of the F_code variable, a larger F_code value may be output as the correlation is lower.

In this embodiment, the F_code detector 3
5, it is assumed that a motion vector and a prediction vector are input. However, the present invention is not limited to this. Only a motion vector is input, a prediction vector is detected inside the F_code detector 35, and the motion vector and the prediction vector are detected. The correlation with the vector may be obtained.

When detecting the correlation with the prediction vector, it is also possible to input only the motion vector to the F_code detector 35 to approximate the correlation between the prediction vector and the motion vector.

Further, not only the motion vector but also the input image itself input to the image coding apparatus 1 and the output data of the adder 37 are output to the F_code detector 35 by the F_code detector 35.
It may be used for calculating a correlation between a motion vector and a prediction vector.

In addition to the image coding method based on MPEG, the F_code is used to divide the motion vector into a variable length coding section (schematic section) and a fixed length coding section (detailed section) to perform coding. It is widely applicable to the image compression method to be performed.

[0065]

An image coding apparatus according to the present invention detects a motion vector from input image processing information and divides the motion vector into a variable length coding section and a fixed length coding section for coding. A flag value for detecting a correlation value of a motion vector in the image based on the image processing information, and determining a flag value representing a coding range of the motion vector using the correlation value and the maximum value of the motion vector in the image A determining means;

Here, in the image coding apparatus according to the present invention, the flag value determining means calculates the correlation value from the motion vector and the prediction vector, and when the correlation value is concentrated near zero, the fixed-length code The flag value is determined based on the maximum value of the motion vector so that the ratio of the motion vector decreases, and when the correlation value is dispersed from near zero, the motion vector The flag value is determined based on the maximum value.

Therefore, the image coding apparatus according to the present invention prevents the code length of the coded code in the variable length coding unit from being increased even when the prediction vector cannot be calculated with sufficient accuracy. It is possible to prevent the image quality of the encoded image from deteriorating due to an increase in the amount of generated bits of the motion vector.

An image encoding method according to the present invention comprises the steps of: detecting a motion vector from input image processing information; dividing the motion vector into a variable length encoding unit and a fixed length encoding unit; A flag value determining step of detecting a correlation value of a motion vector in the image based on the image processing information and determining a flag value representing a coding range of the motion vector using the correlation value and the maximum value of the motion vector in the image Is provided.

Here, in the image encoding method according to the present invention, in the flag value determining step, the correlation value is calculated from the motion vector and the prediction vector, and when the correlation value is concentrated near zero, the fixed-length code The flag value is determined based on the maximum value of the motion vector so as to reduce the ratio of the motion vector, and when the correlation value is dispersed from near zero, the maximum value of the motion vector is Determine the flag value based on the value.

Therefore, according to the image encoding method of the present invention, even when the prediction vector cannot be calculated with sufficient accuracy, it is possible to prevent the code length of the encoded code in the variable length encoding unit from becoming large. Thus, it is possible to prevent the image quality of the encoded image from deteriorating due to an increase in the amount of generated bits of the motion vector.

[Brief description of the drawings]

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

FIG. 2 is a block diagram illustrating a configuration of a VLC unit in the image encoding device shown as an embodiment of the present invention.

[Explanation of symbols]

1 image encoding device, 11 first frame memory, 1
2 motion vector detector, 13 first computing unit, 14
DCT converter, 15 quantizer, 16 VLC, 17
Buffer, 18 inverse quantizer, 19 IDCT unit, 20
Second arithmetic unit, 21 Second frame memory, 22
Motion compensator, 31 MV buffer, 32 motion vector decomposer, 33 motion vector rough part encoder, 34 motion vector detail part encoder, 35 f_code detector, 36
Prediction vector generator, 37 adder

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5C059 KK22 MA00 MA04 MA14 MA23 MC11 ME01 ME13 NN01 NN28 PP05 PP06 PP07 SS06 TA25 TA46 TB07 TC03 TC12 TD04 UA02 UA31 5J064 AA02 BA01 BA09 BA16 BB03 BC01 BC16 BD02

Claims (10)

[Claims] An image encoding apparatus for detecting a motion vector from input image processing information and dividing and encoding the motion vector into a variable-length encoding unit and a fixed-length encoding unit. Detecting a correlation value of the motion vector, and using a flag value determining means for determining a flag value representing a coding range of the motion vector using the correlation value and the maximum value of the motion vector in the image, Image encoding device. 2. The method according to claim 1, wherein the flag value determination unit detects the correlation value using a prediction vector and a motion vector used when coding the motion vector.
The image encoding device according to claim 1. 3. The flag value determining means calculates the correlation value from the motion vector and the prediction vector, and when the correlation value is concentrated near zero, the ratio of the fixed length encoding unit is small. 3. The image coding apparatus according to claim 2, wherein the flag value is determined based on the maximum value of the motion vector. 4. The flag value determining means calculates the correlation value from the motion vector and the prediction vector. If the correlation value is dispersed from near zero, the flag value determining means determines the ratio of the variable length coding unit. 3. The image encoding apparatus according to claim 2, wherein the flag value is determined based on the maximum value of the motion vector so as to reduce. 5. The method according to claim 1, wherein the flag value determining means calculates the correlation value using an image based on the input image processing information and a predicted image generated based on a motion vector. The image encoding device according to claim 1. 6. An image encoding method for detecting a motion vector from input image processing information, dividing the motion vector into a variable-length encoding unit and a fixed-length encoding unit, and encoding the image. Detecting a correlation value of the motion vector of the motion vector, the flag value determining step of determining a flag value representing a coding range of the motion vector using the correlation value and the maximum value of the motion vector in the image, Image encoding method. 7. The image encoding method according to claim 6, wherein in the flag value determining step, the correlation value is detected using a prediction vector and a motion vector used when encoding the motion vector. . 8. In the flag value determining step, the correlation value is calculated from the motion vector and the prediction vector, and when the correlation value is concentrated near zero, the ratio of the fixed length encoding unit is small. The image encoding method according to claim 7, wherein the flag value is determined based on the maximum value of the motion vector. 9. In the flag value determining step, the correlation value is calculated from the motion vector and the prediction vector, and when the correlation value is dispersed from near zero, the ratio of the variable length coding unit 8. The image encoding method according to claim 7, wherein the flag value is determined based on the maximum value of the motion vector so that the number of motion vectors decreases. 10. The flag value determining step includes calculating the correlation value using an image based on the input image processing information and a predicted image generated based on a motion vector. 7. The image coding method according to 6.