JPH0662388A - Dynamic image compressor - Google Patents

Abstract

PURPOSE:To obtain a high predicting efficiency for any picture, and to improve a picture quality and a compressibility. CONSTITUTION:A dynamic image compressor is equipped with a predicted structure deciding device 2 which judges which correlation is stronger; a field correction or time correlation for the picture based on picture data imparted to decide a predicted structure, and decides the predicted structure by selecting the predicted structure from among a predicted picture group stored in a predicted structure group memory(PM) 3, predicted structure group memory(PM) 3 which stores plural predicted structure groups set based on the field correlation and the time correlation, and compressor 4 which compresses picture data according to the predicted structure selected by the predicted structure deciding device 2. Then, the predicted structure in which an error can be reduced is selected by predicting a P4 picture predicted close in time from an 12 picture and a P3 picture whose field is the same on the same condition, and the picture data are compressed according to the predicted structure.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture compression apparatus used for moving picture compression processing and the like, and more particularly to a moving picture compression apparatus with prediction in the time axis direction.

[0002]

2. Description of the Related Art As an international standard for image compression, JPEG (Jo
int Photograghic Expert Group) and MPEG (Moving
There is a Picture Expert Group).

MPEG is MPEGI, MPEGII, M
A three-level draft of PEGIII is under consideration. MP
EGI aims to compress a moving image that can be transmitted through a communication line of 1.5 Mbps, and is considered to be used mainly in videophones and videoconferences. MPEG I
Then, the current NTSC format video image is 320 x 24
It is treated as a resolution of 0 pixel, and data of only one field is used out of two fields constituting one frame. In MPEGII, compression is a goal that can be transmitted through a 10 Mbps communication line, and moving image transmission by ISDN and digital video are targets. And
MPEGIII is targeted for next-generation televisions such as HDTV.

The feature of MPEG is that DCT (Discrete Cos
ine Transform: Discrete cosine transform) is used to perform inter-frame prediction processing for compression in the time axis direction in addition to still image compression. Random access of frames and fast-forwarding are prerequisites for moving image compression. It has been mentioned that playback and rewind playback (reverse direction) can be performed. Therefore, inter-frame prediction in MPEG employs both forward and backward directions. MPEG
However, basically, MC (motion compensation) + DCT is used. The block size for motion compensation is 16 × 16.
(However, there is also an 8 × 8 mode), DCT is performed on 8 × 8 blocks. Also, this motion compensation is performed with 1/2 pixel precision. Motion compensation with 1/2 pixel accuracy is performed not only by checking the position shifted in pixel units on the reference frame used for prediction, but also by generating the position between pixels by interpolation and performing matching.

The biggest difference from the normal motion compensation + DCT is the motion compensation prediction based on the periodic intra-frame coded frame. 8 and 9 are diagrams showing a prediction structure of a screen at the time of MPEG moving picture coding, and a quadrangle in the drawing means an image (picture) one by one which means a frame of a moving picture.
The arrow extending from the frame indicates that the root frame is used for prediction.

The above pictures are classified into the following types according to the encoding method.

I-picture (Intra-coded picture) When encoded, only the information closed in one image is used. In other words, when decoding, an image can be reconstructed using only the information of the I picture itself. Actually, the DCT is encoded as it is without taking the difference. This encoding method is generally inefficient, but random access and high-speed reproduction are possible by inserting this encoding everywhere and decoding only the I picture. Furthermore, I picture is decoded and stored in memory,
Reverse reading can be performed by repeating reading in the reverse direction.

P picture (Predictive-coded pictur
e: Forward predictive encoded image) P picture is a predictive image (image that serves as a reference for taking a difference)
As I, which was previously decoded in time at the input
Use picture or P-picture. That is, as shown in FIG. 9, it is a frame predicted in one direction from the past. In actuality, it is possible to select, in macroblock units, whichever is more efficient, whether the difference from the motion-compensated predicted image is coded or the difference is not taken (intra-coding).

B-picture (Bidirectionally predicti
ve-coded picture: bidirectional predictive-coded image) A B-picture is a predicted image, which is a temporally preceding and already decoded I picture or P picture, and a temporally posterior already decoded I picture, or Three types of P-pictures and interpolated images made from both are used. Here, in the case of an interpolation frame, prediction is performed from both directions, but there are roughly three prediction modes for motion compensation. These are forward motion compensation that predicts the present from the past, backward motion compensation that predicts the present from the future, and interpolation motion compensation that predicts the present from both the past and the future. The forward motion compensation and the backward motion compensation are the same processes as the normal motion compensation (MC) in that they match a block read from one reference frame. In the interpolation motion compensation, the blocks read from the two reference frames are weighted and combined in consideration of the time distance between the current frame and the reference frame to obtain a prediction signal.

The most efficient one of the above three types of difference encoding and intra-encoding after motion compensation can be selected in macroblock units.

On the other hand, the I picture, P picture and B picture
GOP (Group of
A picture) is composed of one or more I-pictures and zero or more non-I-pictures. In order to encode or decode a B picture, an I picture or P picture that is a temporally backward image that becomes a predicted image of the B picture must be encoded first. , P pictures, and B pictures need to be in a predetermined order, but the intervals between I pictures and the intervals between P pictures are arbitrary and may change within the GOP.

[0012]

As described above, in the conventional temporal direction prediction for a moving image, one-sided prediction from the immediately preceding image shown in FIG. 6 and bidirectional prediction from past and future images shown in FIG. Etc.

It is generally considered that the bidirectional prediction shown in FIG. 7 is effective for moving picture compression in a storage medium (MPEG) system.
An image serving as a prediction source of a picture is an image (I picture) that can be reproduced by the image alone (I picture) or an image (P picture) based on one-sided prediction other than a B picture. However, since full moving image data has the field structure described above, if the image to be compressed is not in the same field as the prediction source and there is no vertical change over time, the prediction efficiency will deteriorate considerably. There is a drawback that it will end up. That is, for example, in a television image, one field is displayed every 1/60 seconds, and two fields (odd field and even field) form one frame, and there is a shift of one dot in the spatial axis direction.
Therefore, the ordinates of adjacent images do not match, and the closest image is not always the most similar when trying to make a prediction. May become extremely worse.

By the way, in the P picture described above, it is considered that the prediction efficiency is not so high in many cases in terms of the amount of time elapsed from the prediction source image. If the P picture is in a field different from that of the prediction source, the prediction efficiency will be significantly reduced depending on the image, which may affect other images. However, if the P picture is in the same field as the prediction source, one I
A series of all P pictures with a picture as a prediction source may be the same field, and in a bidirectional prediction screen (B picture) with those I pictures and P pictures as prediction sources, a field different from both the prediction sources may occur. However, there is a drawback that the prediction efficiency is reduced depending on the image.

As described above, since the conventional predictive structure based on I-pictures, P-pictures, etc. is uniquely determined by the designer based on the time correlation, the image to be compressed is not in the same field as the predictor and the time elapses. In addition, when there is no change in the vertical direction, the prediction efficiency may be considerably deteriorated. On the contrary, even when the P picture is in the same field as the prediction source, the prediction may not be possible depending on the image for the above reason. This may lead to a decrease in efficiency.

Therefore, an object of the present invention is to provide a moving picture compression apparatus capable of obtaining high prediction efficiency for any picture.

[0017]

The invention according to claim 1 is
To achieve the above object, a prediction structure storage unit that stores a plurality of prediction structures configured from a plurality of prediction images, and a prediction structure that selects a prediction structure stored in the prediction structure storage unit based on image data to be encoded The selection means and the data compression means for compressing the image data according to the prediction structure selected by the prediction structure selection means are provided.

The plurality of prediction structures stored in the prediction structure storage means are, for example, as described in claim 2.
It may be configured to include a prediction structure used when the field correlation is high and a prediction structure used when the time correlation is high, and the prediction structure is defined in, for example, claim 3. As described, a plurality of images including a coded image that can be independently coded, a one-sided prediction image that predicts the present from the past, and a bidirectional prediction image that predicts the present from both past and future directions. The structure may be configured such that prediction is performed by using each of the images as a prediction source image in a predetermined order. In addition, the prediction structure has a first coded image that can be coded independently in one field obtained by dividing a frame into two fields, an odd field and an even field, as described in claim 4. , The other field may be configured to include a second coded image that can be coded independently, and the prediction structure may be, for example, as described in claim 6. In addition, inter-frame prediction processing for compression in the time axis direction may be performed. Further, the bidirectional predicted image is, for example, as described in claim 5, two adjacent images located between the coded image or the one-sided predicted image and the coded image or the one-sided predicted image. You may make it comprise.

Further, the predictive structure selecting means determines, for example, which of image correlation to be coded is stronger, field correlation and time correlation, as described in claim 7, A prediction structure having a strong correlation may be selected from a plurality of prediction structures stored in the prediction structure storage means based on the determination result, and the prediction structure selection means may be, for example, As described in, the prediction error when predicting the one-sided prediction image that is temporally close when the coded image is used as a prediction source image, and the coded image is the source of the prediction. It is also possible to determine the correlation by comparing with the prediction error when the one-sided prediction image having the same field when used as the image of 1.

[0020]

The operation of the means of the present invention is as follows.

Claims 1, 2, 3, 4, 5, 6, 7 and 8
In the described invention, it is assumed that the prediction structure storage means stores a prediction structure used when the field correlation is high and a prediction structure used when the time correlation is high.

In this state, when the image data to be encoded is input to the prediction structure selecting means, the prediction structure selecting means determines which of the field correlation and the time correlation the image data has, and A prediction structure having a strong correlation is selected from the plurality of prediction structures stored in the prediction structure storage means based on the determination result. And
The image data is compressed according to the prediction structure selected by the prediction structure selecting means.

Therefore, high prediction efficiency can be obtained in any image.

[0024]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below with reference to the drawings.

Description of Principle According to the present invention, there are two states, that is, the image to be compressed is not in the same field as the prediction source, and there is no change in the vertical direction with the passage of time, and if it is in the same field as the prediction source. Focusing on, the two images to be compressed are prepared, and by changing the prediction structure by selecting the prediction structure that uses the higher correlation between the field correlation and the time correlation according to the image information, To make up for such drawbacks,
This is to obtain high prediction efficiency in any image.

FIG. 1 is a functional block diagram of a moving picture compression apparatus according to the present invention. In the figure, a white arrow indicates a data flow and an arrow indicates a control flow.

In FIG. 1, the moving image compression apparatus is a frame memory (F) for storing original image data to be data-compressed.
M) 1 and which of the field correlation (see FIG. 5) and the time correlation (see FIG. 6) the image has based on the image data read from the frame memory (FM) 1 to determine the prediction structure. Is set, based on the field correlation and the time correlation, and the prediction structure determination device 2 that determines the prediction structure by selecting the prediction structure from the prediction image group stored in the prediction structure group memory (PM) 3. A prediction structure group memory (PM) 3 that stores a plurality of prediction structure groups, and a compressor 4 that compresses the image data read from the frame memory (FM) 1 according to the prediction structure selected by the prediction structure determination device 2. The control device 5 controls the operation of each of the above parts.

In the above structure, first, the image data read from the frame memory (FM) 1 is output to the prediction structure determination device 2. The prediction structure determination device 2 determines which one of the image field correlation and the time correlation is stronger based on the image data given for determining the prediction structure, and determines from the prediction structure group memory (PM) 3 Select the predictive structure to determine the predictive structure. The result of determination of the prediction structure is sent to the control device 5, and the control device 5 gives the compressor 4 information from which position and how to predict. The compressor 4 compresses the image data read from the frame memory (FM) 1 according to the prediction structure corresponding to the determined correlation, and outputs it as compressed image data. The above procedure is repeated for each image group as shown in FIG. 4 described later.

As described above, the moving picture compression apparatus based on the above explanation of the principle determines which of the field correlation and the time correlation the image has, based on the image data given for determining the prediction structure. Prediction structure determination device 2 for determining and selecting a prediction structure from a prediction image group stored in the prediction structure group memory (PM) 3, and a plurality of predictions set based on field correlation and time correlation A predictive structure group memory (PM) 3 for storing a structure group and a compressor 4 for compressing image data according to the predictive structure selected by the predictive structure determining device 2 are provided, and a temporally predicted P picture is predicted from an I picture. P-pictures with the same field are predicted under the same conditions, a prediction structure with a small error is selected, and the image data is compressed with this prediction structure. Oite also it is possible to obtain high prediction efficiency. Further, since the prediction efficiency can be increased, the image quality can be improved when the compression rate is the same, and the compression rate can be improved when the image quality is the same.

Embodiment FIG. 2 to FIG. 7 are views showing an embodiment of a moving picture compression apparatus according to the present invention.

First, the structure will be described. FIG. 2 is a block diagram of a moving image compression apparatus. In this figure, an encoder of the moving image compression apparatus outputs an image mode, a prediction mode, a motion vector and various control signals to control the entire system. The controller 30, an image memory 31 for storing image data to be data-compressed, a subtracter 32 for subtracting the prediction result of the motion-compensated inter-frame prediction process from the image data read from the image memory 31, and a subtractor 32 for subtraction. DC which performs DCT operation on the image data while outputting the image data to the controller 30.
The T calculation unit 33, the quantization unit 34 that quantizes the output data of the DCT calculation within a certain error range according to the quantization width determined by the controller 30, and the image data quantized by the quantization unit 34. On the other hand, in addition to image data, various block attribute signals are variable-length coded, and then VLC (Variable) which is multiplexed into a code string of a predetermined data structure.
Length Code (variable length coding) 35 and a buffer 36 for smoothing fluctuating information generation to a constant rate.
And a motion-compensated inter-frame prediction unit 37 that performs motion-compensated prediction based on periodic intra-frame coded frames,
It is composed by.

The motion-compensated inter-frame prediction unit 37 dequantizes the image data quantized by the quantization unit 34 and the dequantization unit 38, and the dequantization unit 38 restores the unquantized image data. Inverse DCT (IDC
T) According to the image mode and prediction mode from the controller 30, the IDCT calculation unit 39 that performs the calculation, the adder 40 that adds motion compensation to the data returned to the image data before the DCT processing by the IDCT calculation unit 39. It is composed of switches 41, 42 and 43 for switching signal paths, and predictors 44 and 45 for performing motion compensation prediction based on motion vectors calculated by the controller 30 (see FIG. 7).

The image memory 31 corresponds to the frame memory (FM) 1 in the functional block diagram of FIG.
The CT calculator 33 and the quantizer 34 generally correspond to the compressor 4 of FIG. The controller 30 generally corresponds to the prediction structure determination device 2, the prediction structure group memory (PM) 3 and the control device 5 of FIG. 1, and the controller 30 uses the built-in prediction structure group memory (PM). Then, the predictive structure determining process is executed, and the moving image compression control of the entire system is performed based on the determined predictive structure.

Further, the decoder of the moving picture compression apparatus performs an operation reverse to that of the above-mentioned encoder, and more specifically,
As shown in FIG. 2, a buffer 46 for smoothing fluctuating information generation to a constant rate, and image data to be decoded stored in the buffer 46, which is the reverse of the processing of the video multiplex encoding unit (VLC) 35. Inverse video multiplex decoding unit (VLC-1) 4 for performing decoding by performing
7 and VL according to the quantization width determined by VLC-147
An inverse quantizer 48 for inverse quantizing the output of C-147;
Inverse DCT is performed on the data inversely quantized by the inverse quantizer 48.
IDCT calculation unit 49 for performing calculation, and IDCT calculation unit 49
VLC-14 and an adder 50 for adding the prediction result to the output of
It is composed of switches 51, 52 and 53 for switching signal paths according to the image mode and prediction mode from 7 and predictors 54 and 55 for performing motion compensation prediction based on the motion vector calculated by VLC-147.

FIG. 4 is a diagram for explaining an image group used in the above-mentioned moving image compression apparatus. In the figure, a rectangle is a screen, and I, P and B are I picture, P picture and B respectively.
Shows a picture.

In general, in compression involving temporal prediction, an image as a prediction source cannot be expanded from the compressed data unless the image has been reproduced. Image compression, which is currently considered to be effective, is a compressed image that does not involve time prediction in order to deal with special playback (intermediate playback, high-speed playback, etc.) during image expansion, and image distortion due to error accumulation. (I picture) exists. In this case, the image data from one I picture to the next I picture is not affected by the image information outside them, so that those screen groups should be processed as one group (image group) as shown in FIG. You can

FIGS. 5 and 6 are diagrams showing a prediction structure based on two correlations. FIG. 5 shows a prediction structure when the field correlation is high, and FIG. 6 shows a prediction structure when the time correlation is high. . The prediction structure group memory (PM) 3 for storing the plurality of prediction structure groups shown in FIG. 1 is composed of a plurality of prediction structures as shown in FIGS. 5 and 6.

The predictive structure of the main moving image compression apparatus is
The following GOP (Group Of Picture) is configured to change the prediction structure according to the image information.

First, the number of B pictures predicted bidirectionally in the past and the future is fixed at, for example, "2". That is, basically, any number of B pictures may exist between P pictures and I pictures, but in the present embodiment, by fixing this B picture to 2, one of the two B pictures must be The vertical axis of the spatial axis exists between the adjacent P picture or I picture, and therefore, the other of the B pictures is always the temporally closest image. That is, a B picture is predicted from a P picture (this P picture is predicted from an I picture), but by fixing the number of B pictures to 2 as shown in FIGS. The picture and I picture (or P picture) have a closed relationship,
It can be seen as something independent.
In other words, by setting the number of B pictures to 2,
The pictures can be independent and unaffected.

Next, the P picture predicted from one side is determined as follows. A column of the frame memory including one or more I pictures is called a GOP. As described above, in this GOP, two B pictures are taken to obtain a certain GOP.
The number of screens between I-pictures is adjusted so that one I-picture and the next I-picture are different fields. Specifically, FIG. 4 and FIG.
As shown in FIG. 6, if the number of P pictures between the I1 picture and the next I2 picture is even (4 in this embodiment), the I1 picture and the I2 picture are switched so as to take different fields. become.

In this case, since the P picture having the above prediction structure correlates only with the image on one side that is a little distant from the other, whether the P picture image is temporally closer has less prediction error, or the location It is determined whether the prediction error is smaller when the (fields) are the same, and the optimum prediction structure is selected from the prediction structures of FIGS. 5 and 6 based on the determination result. That is, as shown in FIG. 5 and FIG.
The moving image compression apparatus has a prediction structure when the field correlation is high and a prediction structure when the time correlation is high, and selects either one according to the image information.

As the selection method, P3 picture and P4 picture which can be predicted from I2 picture are predicted under the same conditions, respectively, and one having a smaller error in the prediction structure is selected. For example, when the time correlation is stronger, the prediction structure is switched so that the prediction is performed from the closest time point and when the field (location) correlation is strong, the prediction can be performed from the same field. In the field correlation shown in FIG. 5, each P picture is predicted from an image in the same field, and in the time field shown in FIG. 6, each P picture is predicted from an image that is temporally close. In addition, the B picture is configured such that one side is an image that is close in time and the other side is an image in the same field.

Next, the operation of this embodiment will be described.

FIG. 7 is a flowchart of the predictive structure determining process executed by the controller 30, which corresponds to the process in the predictive structure determining device 2 of FIG.

First, in step S1, I2 picture to P4
The picture is predicted, and the I2 picture is converted to the P3 in step S2.
Predict pictures. That is, a P that is temporally closer to the next I2 picture whose field is different from that of the I1 picture.
A P4 picture to be a picture is predicted (see FIG. 6), and a P3 picture having the same P picture in the same field is predicted from the same I2 picture (see FIG. 5).

Then, in step S3, it is determined whether or not the prediction of the P4 picture has a smaller error than the prediction of the P3 picture. In this case, the selection method predicts the P3 picture and the P4 picture that can be predicted from the I2 picture under the same conditions, and selects the one having the smaller error in the prediction structure. Here, in the field correlation, each P picture is predicted from images that are temporally close. A B picture is an image in which one side is temporally close and the other is an image in the same field.

When the error of the P4 picture is smaller than that of the P3 picture, the time correlation is stronger than that of the P3 picture. Therefore, the time correlation prediction structure shown in FIG. After processing, P
When the error is smaller in predicting 3 pictures than in predicting P4 pictures, the field correlation is stronger, so that the field correlation prediction structure shown in FIG. 5 is adopted in step S5 and the processing of this flow ends. .

More specifically, when the prediction structures shown in FIGS. 5 and 6 are prepared, the prediction structure can be determined by the procedure shown in the flow of FIG.

First, from I2 picture to P4 picture and P
Prediction to 3 pictures is prediction including motion compensation (MC), and is performed in the same manner as normal moving picture prediction. By comparing the total of the prediction errors, either P4 picture or P3 picture is predicted. Is closer to the image of the I2 picture, and if the image of the P4 picture is closer (that is, if the prediction error is smaller), the temporal correlation structure of FIG.
If the P3 picture image is closer, the field correlation structure of FIG. 6 is selected. If information indicating which is selected is given for each image group in FIG. 4, normal operation can be performed at the time of decoding. It should be noted that this selection information can be realized by expanding the GOP header by 1 bit in the MPEGI format.

As described above, the controller 30 of the moving picture compression apparatus according to the present embodiment predicts the P4 picture that is temporally close to the I2 picture and the P3 picture having the same field under the same conditions. Since a prediction structure with a small error is selected and the image data is compressed with the prediction structure, high prediction efficiency can be obtained for any image. Further, since the prediction efficiency can be increased, the image quality can be improved when the compression rate is the same, and the compression rate can be improved when the image quality is the same.

In this embodiment, a plurality of prediction structures are
There are two prediction structures, a prediction structure used when the field correlation is high and a prediction structure used when the time correlation is high, but the present invention is not limited to this, and is suitable for each correlation (field correlation, time correlation). Any structure can be applied as long as it is a prediction structure.

Further, in the present embodiment, which correlation between the image data to be coded is stronger between the field correlation and the time correlation is a prediction error when the P4 picture temporally close to the I2 picture is predicted. , The correlation is determined by comparing with the prediction error when the P3 picture with the same field is predicted, but any correlation can be determined, for example, the amount of movement of the image in the vertical direction. May be used.

The above-described prediction structure is an example, and the number of P pictures and B pictures, positions, prediction directions, etc. are not limited to those in the above embodiments.

Further, in the present embodiment, the moving image compression apparatus is M
This is an example applied to a moving image compression apparatus based on the PEG algorithm, but needless to say, the present invention is not limited to this and can be applied to all apparatuses as long as they perform encoding based on a prediction structure.

Further, it goes without saying that the number and types of circuits and members constituting the above-mentioned moving picture compression apparatus are not limited to those in the above-mentioned embodiment, but may be realized by software (for example, C language). .

[0056]

According to the first, second, third, fourth, fifth, sixth, seventh and eighth aspects of the present invention, a prediction structure storage means for storing a plurality of prediction structures each composed of a plurality of prediction images, and a code A prediction structure selecting means for selecting a prediction structure stored in the prediction structure storing means based on image data to be converted; and a data compressing means for compressing the image data according to the prediction structure selected by the prediction structure selecting means. Therefore, it is possible to obtain high prediction efficiency for any image, and it is possible to improve image quality and compression rate.

[Brief description of drawings]

FIG. 1 is a functional block diagram of a moving image compression apparatus.

FIG. 2 is a diagram showing a block configuration of an encoder of a moving image compression apparatus.

FIG. 3 is a diagram showing a block configuration of a decoder of a moving image compression apparatus.

FIG. 4 is a diagram showing an image group of a moving image compression apparatus.

FIG. 5 is a diagram showing a prediction structure of a moving image compression apparatus.

FIG. 6 is a diagram showing a prediction structure of a moving image compression apparatus.

FIG. 7 is a flowchart showing a predictive structure determination process of the moving image compression apparatus.

FIG. 8 is a diagram showing a prediction structure of a screen of the moving image compression apparatus.

FIG. 9 is a diagram showing a prediction structure of a screen of the moving image compression apparatus.

[Explanation of symbols]

 1 Frame Memory (FM) 2 Prediction Structure Determining Device 3 Prediction Structure Group Memory (PM) 4 Compressor 5 Controller 30 Controller 31 Image Memory 32 Subtractor 33 DCT Calculator 34 Quantizer 35 VLC 36, 46 Buffer 37 Motion Compensation Inter-frame predictor 38,48 Dequantizer 39,49 IDCT calculator 40,50 Adder 41,42,43,51,52,53 Switch 44,45,54,55 Predictor 47 Inverse VLC

Claims (8)

[Claims] 1. A prediction structure storage unit for storing a plurality of prediction structures each composed of a plurality of prediction images, and a prediction structure selection unit for selecting a prediction structure stored in the prediction structure storage unit based on image data to be encoded. And a data compression unit that compresses image data according to the prediction structure selected by the prediction structure selection unit. 2. A plurality of prediction structures stored in the prediction structure storage means include a prediction structure used when field correlation is high and a prediction structure used when time correlation is high. The moving picture compression apparatus according to claim 1, wherein 3. The prediction structure includes a plurality of coded images that can be coded independently, a one-sided prediction image that predicts the present from the past, and a bidirectional prediction image that predicts the present from both past and future directions. 3. The moving image according to claim 1, wherein the moving image has a structure configured to perform prediction by using each image as a prediction source image in a predetermined order. Image compression device. 4. The prediction structure has a first coded image that can be independently coded in one field obtained by dividing a frame into an odd field and an even field, and independently in the other field. 4. The moving image compression apparatus according to claim 1, wherein the moving image compression apparatus is configured to include a second encoded image that can be encoded. 5. The bidirectionally predicted image is two adjacent images located between the coded image or the one-sided predicted image and the coded image or the one-sided predicted image. 3. The moving image compression apparatus according to item 3. 6. The predictive structure performs interframe predictive processing for compression in a time axis direction, claim 1, claim 2, claim 3, claim 4 or claim 6. 5. The moving image compression apparatus according to any one of 5 above. 7. The predictive structure selecting means determines which of image correlation to be coded is stronger, field correlation or time correlation, and the predictive structure storing means stores it in the predictive structure storage means based on the determination result. Of the stored prediction structures,
The moving picture compression apparatus according to claim 1, wherein a prediction structure having a strong correlation is selected. 8. The prediction structure selecting means, a prediction error when predicting the one-sided predicted image that is temporally close when predicted using the coded image as a prediction source image,
When the coded image is used as an original image for prediction, the correlation is determined by comparing with a prediction error when the one-sided predicted image having the same field is predicted. The moving image compression apparatus according to claim 1.