JP2003274411A - Image encoding method and image decoding method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To enhance a conventional image encoding method and a conventional image decoding method wherein a plurality of frames are used for inter-frame prediction reference frames and to enhance the compression rate more than that of the conventional methods. <P>SOLUTION: In the image encoding method for including a bidirectional prediction frame for inter-frame prediction from frames whose display times are before and after that of an encoding object frame, a forward prediction frame making inter-frame prediction from only a frame whose display time is before that of the encoding object frame, and an in-plane prediction frame not making the inter-frame prediction as prediction frames, when the encoding object frame is the forward prediction frame or the bidirectional prediction frame, two frames whose display times are earlier than that of the encoding object frame are used for the reference frames to produce prediction pixel values for the inner-frame prediction by pixel interpolation. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of encoding and decoding an image signal, and a recording medium having a program for implementing the same implemented by software.

[0002]

2. Description of the Related Art In recent years, with the development of multimedia applications, it has become common to uniformly handle information of all media such as images, voices and texts. At this time, it becomes possible to handle media in a unified manner by digitizing all media. However, since a digitized image has a huge amount of data, an image information compression technique is indispensable for storage and transmission. On the other hand, standardization of compression technology is also important in order to interoperate compressed image data. Image compression technology standards include ITU (International Telecommunication Union Telecommunication Standardization Sector) H.261, H.263, and ISO (International Organization for Standardization) MPEG (Moving Picture Experts Gro).
up) -1, MPEG-2, MPEG-4 and so on.

In H.26L, which is currently being standardized by the ITU, a method of obtaining a prediction pixel by interpolating pixels using the two frames immediately before the encoding target frame as reference frames is being studied.

FIG. 18 is a block diagram showing the structure of a conventional image coding apparatus. The image coding device uses the image signal Img
Is divided into rectangles of a predetermined size (hereinafter referred to as blocks), and processing is performed for each block. The subtractor Sub is a predictive image signal Pre from the image signal Img input to the image encoding device.
Subtract d and output as residual signal Res. The image coding means ImgEnc inputs the residual signal Res, performs image coding processing such as DCT conversion and quantization, and outputs a coded residual signal ERes including quantized DCT coefficients. The image decoding means ImgDec inputs the encoded residual signal ERes and performs inverse quantization /
Image decoding processing such as inverse DCT conversion is performed to obtain a decoded residual signal
Output DRes. The adder Add adds the decoded residual signal DRes and the predicted image signal Pred, and outputs the reconstructed image signal Recon. The reconstructed image signal Recon, which may be referred to in the subsequent inter-frame prediction, is stored in the multi-frame buffer MFrmBuf. The motion estimator ME0 receives the image signal Img and the reference frame signal RefFrm, detects a motion in a block, or a region unit obtained by further dividing the block, and a motion information MV and a reference frame identifier indicating the reference frame from the detection result. It outputs RefFrmID and the block prediction type PredType indicating the block prediction type.

The prediction indicated by the block prediction type PredType includes in-plane prediction that does not use interframe prediction and image signal Im.
There is a forward prediction that predicts from the frame of the time before g. Blocks that have been subjected to in-plane prediction and forward prediction are referred to as in-plane prediction block and forward prediction block, respectively. Here, a frame including only the intra prediction block is referred to as an intra prediction frame, and a frame including the forward prediction block and the intra prediction block is referred to as a forward prediction frame.
Further, forward prediction includes a method of obtaining a predicted image from one frame and a method of obtaining a predicted image from two frames by pixel interpolation. A block used to obtain a predicted image by pixel interpolation is called a forward interpolation predicted block. These block prediction types can be switched in block units.

In the case of the forward prediction block, the frame immediately preceding the frame to be coded is selected as the reference frame. In the case of the forward interpolation prediction block, the frame immediately preceding and the frame two frames before the encoding target frame are used as reference frames.

The concept of forward interpolation prediction will be described with reference to FIGS. 19 and 20. FIG. 19 is a conceptual diagram of frame selection at the time of forward interpolation prediction in the conventional image coding method. The horizontal axis indicates time. Frm1, Frm2, Frm3 are frame and block Coding
Blk indicates a forward interpolation prediction block. The current frame to be encoded is P_Frm3. The previous frames Frm2 and 2 obtained from the motion estimation results for the block CodingBlk
The positions of the predicted pixel in the frame Frm1 immediately before are set to block PredBlk1 and block PredBlk2, respectively. The predicted pixel value of block CodingBlk is obtained by pixel interpolation of block PredBlk1 and block PredBlk2. The interpolation method performed by the forward prediction pixel interpolation unit FwPol includes, for example, an average value and extrapolation. Extrapolation is highly effective in predicting when pixel values change linearly with the passage of time such as fade. Fade is image processing in which pixel values change linearly with time, and there are fade-in and fade-out in which pixel values monotonously increase or decrease, and cross-fading, which is bilinear interpolation of images with two pixel values.

FIG. 20 is an explanatory diagram of a pixel value change due to a fade. The horizontal axis represents time, and the vertical axis represents pixel value.
Pt0, Pt1 and Pt2 are blocks CodingBlk and PredBl, respectively.
Pixels with the same position in k2 and PredBlk1 are shown. P0, P1, P2
Indicates the pixel values of the pixels Pt0, Pt1, and Pt2, respectively. The line Pixel_Value_Line in the graph indicates that the pixel value linearly increases due to the fade. At this time, the value of the point Pt0 of the encoding target frame is the points Pt1 and Pt2 of the reference frame.
Can be predicted from the extrapolation calculation of the following equation. P0 = 2 x P1-P2

Therefore, if the forward prediction pixel interpolating means FwPol is used to extrapolate the above equation, the prediction effect against fades is enhanced and the coding efficiency is improved. Also, for images without fade, the number of prediction types is increased by introducing interpolation (average value) instead of extrapolation, and the most suitable prediction method for the block can be selected and the coding efficiency is improved. .

The reference frame identifier RefFrmID indicates one reference frame in the case of a forward prediction block and two reference frames in the case of a forward interpolation prediction block. The multi-frame buffer MFrmBuf is a reference frame identifier RefFr.
The pixel of the block at the position indicated by mID and motion information MV is output as the predicted image signal Pred. If the block prediction type PredType indicates forward interpolation prediction, two image signals of blocks are output, and if it indicates other than forward interpolation prediction, block 1 is output.
Output individual image signals. If the block prediction type PredType indicates forward interpolation prediction, the switch SW0 is switched to the 0 side, the pixel value is interpolated by the forward prediction pixel interpolating means FwPol, and the predicted image signal Pred is output. Block prediction type
If PredType indicates anything other than forward interpolation prediction, switch
Switch SW0 to 1 side, and multi-frame buffer MFrmBuf
The predicted image signal Pred from is used as it is. When the block prediction type PredType indicates in-plane prediction, the predicted image signal Pred is set to 0 value. Variable length coding means VLC
Encodes the coded residual signal ERes, the motion information MV, and the block prediction type PredType with variable length coding, and outputs the image coded signal BitStrm0.
Output as.

By the processing described above, the image coding apparatus codes the image signal Img and outputs it as the image coded signal BitStrm0.

FIG. 21 is a conceptual diagram of a format of an image coded signal of a conventional image decoding apparatus. The format of FIG. 21 shows the format of the forward interpolation prediction block. Frame is a 1-frame signal, Block is a 1-block signal, and Block prediction type PredTyp
e and the encoded signal of the motion information MV are stored. In the forward interpolation prediction block, information indicating the reference frame is not stored in the image coded signal because it is predetermined that the frame immediately before and the second frame with respect to the encoding target frame are used as the reference frame in advance.

FIG. 22 is a block diagram showing the structure of a conventional image decoding apparatus. Units that perform the same operations as those in the block diagram showing the configuration of the conventional image coding apparatus in FIG. 18 and signals of the same operations are denoted by the same symbols, and description thereof is omitted.

The variable length decoding means VLD is for encoding the coded image signal Bit.
Strm0 is input, variable length decoding is performed, and the coded residual signal ERe is input.
s, motion information MV, and block prediction type PredType are output. The image decoding means ImgDec inputs the encoded residual signal ERes, performs image decoding processing such as inverse quantization / inverse DCT conversion, and outputs a decoded residual signal DRes. The adder Add adds the decoded residual signal DRes and the predicted image Pred to obtain the decoded image signal DI
Output as mg outside the image decoding device. The multi-frame buffer MFrmBuf stores the decoded image signal DImg in the buffer for inter-frame prediction. Reference frame selection means Re
fFrmSel0 selects a reference frame according to the block prediction type PredType, and instructs the reference frame to the multi-frame buffer MFrmBuf by the reference frame identifier RefFrmID indicating the reference frame. If the block prediction type PredType indicates forward interpolation prediction, two block image signals are output, and if block prediction other than forward interpolation prediction is indicated, one block image signal is output. If the block prediction type PredType indicates forward interpolation prediction, the switch SW0 is switched to the 0 side, the pixel value is interpolated by the forward prediction pixel interpolating means FwPol, and the predicted image signal Pred is output. Block prediction type Pr
If edType indicates anything other than forward interpolation prediction, switch SW
The 0 is switched to the 1 side, and the predicted image signal Pred from the multi-frame buffer MFrmBuf is used as it is. When the block prediction type PredType indicates in-plane prediction, the predicted image signal Pred is set to 0 value. The variable length coding means VLC is
Encoded residual signal ERes, motion information MV, block prediction type Pr
edType is variable length coded and output as an image coded signal BitStrm0.

By the processing described above, the image decoding apparatus decodes the image coded signal BitStrm0 and outputs it as the image decoded signal DImg. On the other hand, in image coding, there is also a method of performing prediction not only from a temporally previous frame but also from temporally preceding and following frames. By using temporally previous and subsequent frames for prediction, the number of prediction directions increases, and it becomes possible to select a more optimal prediction direction and the coding efficiency increases.

[0016]

The combination of forward interpolation prediction and bidirectional prediction has not been studied so far. In bidirectional prediction, the order of the frames to be encoded / decoded and the order of the frames to be displayed do not match, so not only the frame whose encoding / decoding time is close is used for prediction as in the past, It is necessary to predict the display order. In addition, the scalability of bidirectional prediction frames, which is an advantage of bidirectional prediction (bidirectional prediction frames compared to other frames, is important if we do not distinguish between cases where bidirectional prediction can be used and cases where bidirectional prediction is not possible in forward interpolation prediction. It is not important). By solving these problems of bidirectional prediction, it is possible to improve the compression rate using bidirectional prediction.

Therefore, the present invention provides an encoding method and a decoding method that combine forward interpolation prediction and bidirectional prediction.

[0018]

[Means for Solving the Problems] First invention (Claim 1)
Is a bidirectional prediction frame that performs inter-frame prediction from frames before and after the display time, and
In an image coding method that includes a forward prediction frame that performs inter-frame prediction only from a frame whose display time is previous, and an intra-prediction frame that does not perform inter-frame prediction, the target frame is a forward prediction frame or a bidirectional When the frame is a predicted frame, the image coding method is characterized in that a predicted pixel value is generated by pixel interpolation using two frames whose display times are earlier than the frame to be coded as reference frames.

According to a second aspect of the present invention (claim 6), as the predictive frame, a bidirectional predictive frame in which interframe prediction is performed from frames before and after display time, and interframe prediction is performed only from a frame in which display time is previous. In the image coding method including the forward prediction frame and the in-plane prediction frame that does not perform inter-frame prediction, when the coding target frame is the forward prediction frame or the bidirectional prediction frame, This is an image decoding method characterized in that a predicted pixel value is generated by pixel interpolation using two frames whose display times are previous as reference frames.

A third aspect of the present invention (claim 11) is an image encoding method for predicting and encoding a moving image between frames, and a pixel whose reference time is two frames whose display time is before the encoding target frame. In the case of inter-frame prediction by interpolation, two reference frames are arbitrarily selected from frames whose display time is earlier than the encoding target frame, and information for identifying the two reference frames is stored in the image encoded signal. The image coding method is characterized in that

A fourth aspect of the present invention (claim 14) is an image decoding method for predicting and decoding a moving image between frames by pixel interpolation using two frames whose display time is before the decoding target frame as reference frames. In the case of inter-frame prediction, the identification information for identifying two reference frames in the input image coded signal is decoded, and the two frames indicated by the identification information are used as reference frames for prediction by pixel interpolation. The image decoding method is characterized in that

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Specific embodiments of the present invention will be described below with reference to the drawings. (First Embodiment) FIG. 1 is a block diagram of an image coding apparatus according to the first embodiment. Units that perform the same operations as those in the block diagram showing the configuration of the conventional image coding apparatus in FIG. 18 and signals of the same operations are denoted by the same symbols, and description thereof is omitted.

The present invention is an encoding method that combines forward interpolation prediction and bidirectional prediction. Therefore, in order to maintain compatibility with the above description, the block prediction type and the frame type are additionally defined as follows. In addition to the in-plane prediction, the forward prediction, and the forward interpolation prediction, the prediction for the block includes the backward prediction that performs the prediction from the frame whose display time is later and the bidirectional prediction that performs the prediction by the pixel interpolation of the two frames before and after.
Let's make it a kind. In-plane prediction block, forward prediction block, forward interpolation prediction block, backward prediction block, and bidirectional prediction block are respectively subjected to in-plane prediction block, forward prediction block, forward interpolation prediction block, backward prediction block, and bidirectional prediction block. Call it a block. A frame in which a block in the screen is composed of an intra prediction block is called an intra prediction frame. A frame in which a block in the screen is composed of an in-plane prediction block, a forward prediction block, and a forward interpolation prediction block is called a forward prediction frame. The block in the screen is the in-plane prediction block,
A frame including a forward prediction block, a forward interpolation prediction block, a backward prediction block, and a bidirectional prediction block is called a bidirectional prediction frame.

The block prediction type PredType indicates a prediction type including in-plane prediction, forward prediction, forward interpolation prediction, backward prediction, and bidirectional prediction for a block. The multi-frame buffer MFrmBuf is 2 if the block prediction type PredType indicates forward interpolation prediction and bidirectional interpolation prediction.
Outputs the image signal of the block at the position indicated by each motion vector for one reference frame for generation of the predicted image signal Pred, and predicts the image signal of the block indicated by the motion vector for one reference frame if other Output for image signal Pred. If the block prediction type PredType indicates forward interpolation prediction, the switch SW is switched to the 0 side, the pixel value is interpolated by the forward prediction pixel interpolating means FwPol, and the predicted image signal Pred is output. Block prediction type Pr
When edType indicates bidirectional prediction, the switch SW is switched to the 2 side, and the bidirectional prediction pixel interpolating means BiPol generates the predicted image signal Pred. Block prediction type PredType
Shows other than forward interpolation prediction and bidirectional prediction, switch SW to 1 side and set the multi-frame buffer MFrmB.
The predicted image signal Pred of uf is used as it is. BiPol is a multi-frame buffer MFrmB
Pixel interpolation is performed using two frames temporally before and after the frame to be encoded in uf as reference frames, and a predicted image image Pred is output. The frame-type forward interpolation prediction reference frame selection unit FwPolRefFrmSel is a unit that selects a reference frame at the time of forward interpolation prediction for each type of frame, that is, a forward prediction frame and a bidirectional prediction frame. The method of selecting the reference frame will be described later. Reference frame selection means for forward interpolation prediction for each frame type PolRefFrmSel is a reference frame identifier for forward interpolation prediction that indicates two reference frames used in forward interpolation prediction.
Notify the motion estimation means ME of FwPolRefFrmID. In the case of the forward interpolation prediction block, the motion estimation means ME performs the motion estimation using the two frames indicated by the forward interpolation prediction reference frame identifier FwPolRefFrmID as reference frames.

Here, a reference frame selection method at the time of forward interpolation prediction in a forward prediction frame will be described. In the image coding method including the bidirectional prediction frame, the bidirectional prediction frame whose display time is temporally close to the previous one should be coded / decoded later than the one prediction frame. Therefore, the forward prediction block of the forward prediction frame performs prediction by only referring to the in-plane prediction frame or the forward prediction frame, and does not perform prediction by referring to the bidirectional prediction frame. In view of this, the present invention provides a restriction that even in the forward interpolation prediction block of the forward prediction frame, the prediction is performed from the in-plane prediction frame or the forward prediction frame, and the prediction with reference to the bidirectional prediction frame is not performed. However, with respect to the encoding target frame, the two temporally closest in-plane prediction frames or forward prediction frames are used.

FIG. 2 is a conceptual diagram of frame selection at the time of forward interpolation prediction in the forward prediction frame according to the first embodiment. P_Frm1, P_Frm4, P_Frm7 are forward prediction frames, B_Frm
2, B_Frm3, B_Frm5, B_Frm6 indicate bidirectional prediction frames, and block CodingBlk indicates forward interpolation prediction blocks. The current frame to be encoded is P_Frm7. Encoding target frame
The block at the position obtained from the result of motion estimation is blocked using the forward prediction frame P_Frm4, which is temporally earlier than P_Frm7 and is temporally closest to P_Frm7, as the reference frame.
Let k1. P_Frm7 is temporally earlier, and the forward predicted frame P_Frm1 that is the second closest from P_Frm7 is used as a reference frame
Block the block at the position obtained from the result of motion estimation
PredBlk2. The predicted pixel value of the block CodingBlk is
It is obtained by pixel interpolation of block PredBlk1 and block PredBlk2.

FIG. 3 is a conceptual diagram showing a multi-frame buffer at the time of forward interpolation prediction in the forward prediction frame according to the first embodiment. In the multi-frame buffer in the figure, it is assumed that the frames are arranged from the bottom to the top in order from the frame with the earliest display time. It is assumed that the frames of P_Frm1, B_Frm3, and P_Frm4 are stored in the multi-frame buffer MFrmBuf at the time of forward interpolation prediction. In MPEG, the bidirectional prediction frame does not become a reference frame, so the bidirectional prediction frame is not stored in the multi-frame buffer MFrmBuf. However, in H.26L, the bidirectional prediction frame also becomes a reference frame, so the bidirectional prediction frame is stored in the multiframe buffer MFrmBuf. Bi-directional predicted frames may be stored. Therefore, the image coding apparatus needs to consider such a case. In the image encoding device of the present invention, during the forward interpolation prediction of the forward prediction frame P_Frm7, the bidirectional prediction frame B_Frm3 is not used even if the display time is near, and the forward prediction frame P_Frm4 and the forward prediction frame P_Frm1 are used. As in the present invention, when inter-frame prediction including forward interpolation prediction of a forward prediction frame is restricted so as not to refer to a bidirectional prediction frame, the image coded signal output by the image coding apparatus of the present invention is Can have scalability. Scalability means that image coded signals of different rates can be acquired from one image coded signal. In the image coded signal of the present invention, the bidirectional prediction frame is not referenced from the forward prediction frame. Therefore, by removing the bidirectional prediction frame from the image coded signal, the image coded signal of the lower bit rate is obtained. Can be obtained. Due to the scalability, one image coded signal can be used for communication channels of different bands. Also, the image decoding apparatus can reduce the processing load without decoding one copy of the bidirectional prediction frame and perform the real-time reproduction when the processing speed is insufficient for the reproduction.

Next, a reference frame selection method at the time of forward interpolation prediction in a bidirectional prediction frame will be described. In the present invention, three methods are shown as reference frame selection methods for forward interpolation prediction in bidirectional prediction frames.

According to the first reference frame selection method, even in the case of forward interpolation prediction in a bidirectional prediction frame, the two closest intra-prediction frames whose display time is before the encoding target frame or forward prediction This is a method in which prediction is performed only from frames, and prediction is not performed from bidirectionally predicted frames. The first reference frame selection method will be described with reference to FIG.

FIG. 4 is a conceptual diagram of the first frame selection method at the time of forward interpolation prediction in the bidirectional prediction frame according to the first embodiment. P_Frm1, P_Frm4, P_Frm7 are forward prediction frames, B_Frm2, B_Frm3, B_Frm5, B_Frm6 are bidirectional prediction frames, and a block CodingBlk is a forward interpolation prediction block. The current frame to be encoded is B_Frm5. A block at a position obtained from the result of motion estimation is set as a block PredBlk1 with a forward prediction frame P_Frm4 temporally closest to B_Frm5 temporally before the encoding target frame B_Frm5 as a reference frame. Before B_Frm5, B_Frm5
The forward predicted frame P_Frm1 that is second closest to the reference frame is used as a reference frame, and the block at the position obtained from the result of the motion estimation is set as a block PredBlk2. Block Coding Bl
The predicted pixel value of k is the block PredBlk1 and block PredBlk.
It is obtained by interpolation of 2 pixels. The first reference frame selection method is advantageous in terms of scalability. Since the bidirectional prediction frame is not referred to by other frames, excluding an arbitrary bidirectional frame does not affect the reproduction of the remaining frames.

In the second reference frame selection method, at the time of forward interpolation prediction in the bidirectional prediction frame, one of the two reference frames has a display time earlier than the encoding target frame, and the display time is closest to the front. Predicted frame, or.
This is a method in which an in-plane prediction frame is used, and the other is a bidirectional prediction frame whose display time is earlier than the encoding target frame and whose display time is the closest. The second reference frame selection method will be described with reference to FIG.

FIG. 5 is a conceptual diagram of the second frame selection method at the time of forward interpolation prediction in the bidirectional prediction frame according to the first embodiment. P_Frm1, P_Frm4, P_Frm7 are forward prediction frames, B_Frm2, B_Frm3, B_Frm5, B_Frm6 are bidirectional prediction frames, and a block CodingBlk is a forward interpolation prediction block. The current frame to be encoded is B_Frm6. A forward prediction frame P_Frm4 whose display time is earlier than the encoding target frame B_Frm6 and whose display time is closest to B_Frm6 is set as a reference frame, and a block at a position obtained from the result of motion estimation is set as a block PredBlk1. The display time is before B_Frm6, and B_
Bidirectional prediction frame B_F with the second closest display time from Frm6
A block at a position obtained from the result of motion estimation is set as a block PredBlk2 using rm5 as a reference frame. block
The predicted pixel value of CodingBlk is obtained by pixel interpolation of block PredBlk1 and block PredBlk2. If there are only in-plane prediction frames or forward prediction frames in the multi-frame buffer MBufFrm, the first reference frame selection method is used. An advantage of the second reference frame selection method is that inter-frame prediction can be performed for a frame to be coded from a frame whose display time is closer than that of the first reference frame selection method. In general, if the inter-frame prediction is performed from the frame whose display time is closer, the coding efficiency becomes higher.

In the third reference frame selection method, at the time of forward interpolation prediction in a bidirectional prediction frame, regardless of the type of the reference frame, the two closest display times before the encoding target frame are set as the reference frames. Is the way to do it. The third reference frame selection method will be described with reference to FIGS. 6 and 7.

FIG. 6 is a conceptual diagram of a third frame selection method at the time of forward interpolation prediction in the bidirectional prediction frame according to the first embodiment. P_Frm3 and P_Frm7 are forward prediction frames, B_F
rm1, B_Frm2, B_Frm4, B_Frm5, B_Frm6 are bidirectional prediction frames, and the block CodingBlk is a forward interpolation prediction block. The current frame to be encoded is B_Frm6. P_Frm
3, B_Frm4, B_Frm5, P_Frm7 are multi-frame buffer MFrm
It is assumed that the frame is stored in Buf. The frame B_Frm6 whose display time is earlier than the encoding target frame B_Frm6 and whose display time is closest to B_Frm6 is the reference frame, and the block at the position obtained from the motion estimation result is the block PredBlk1. . B_Frm6
The block at the position obtained from the motion estimation result is set as the block Pr using the bi-predictive frame B_Frm4, which is earlier in display time and whose display time is the second closest to B_Frm6, as the reference frame.
Call it edBlk2. The predicted pixel value of the block CodingBlk is obtained by pixel interpolation of the block Pred_Blk1 and the block PredBlk2.

FIG. 7 is a conceptual diagram showing a multi-frame buffer at the time of forward interpolation prediction in the bidirectional prediction frame according to the first embodiment. The frames stored in the multi-frame buffer MFrmBuf are P_Frm3, B_Frm4, B_Frm5, P_Frm.
Set to 7. If the frames whose display time is earlier than the encoding target frame B_Frm6 are arranged in the order of being closer to B_Frm6, B_Frm5, B_F
The order is rm4, P_Frm3. Therefore, B_Frm5 and B_Frm4 are used as reference frames. The advantage of the third reference frame selection method is that the effect of extrapolation interpolation on fades can be increased. If the extrapolation interpolation can be performed by the following equation as described in the conventional example, the prediction effect can be enhanced while suppressing the processing amount required for the interpolation to be low. P0 = 2 x P1-P2

However, in the extrapolation of the above equation, it is assumed that the display time interval of a total of three frames of the reference frame and the coded frame is constant. Depending on the number of bidirectionally predicted frames between the in-plane predicted frame or the forward predicted frame, the display time interval between the reference frame and the coded frame may not always be constant. number 3
In the reference frame selection method of, since the two frames immediately before the encoding target frame can be stored in the multi-frame buffer and used for prediction, the reference frame and the encoding frame are displayed under a constant input frame rate. It is possible to keep the time interval constant. Therefore, the third
The reference frame selection method of
If two frames are stored in the multi-frame buffer MFrmBuf, high prediction efficiency with respect to fade can be realized, and coding efficiency is improved.

In the inter-frame prediction including the forward prediction frame and the forward interpolation prediction of the bidirectional prediction frame, if there is an in-plane prediction frame whose display time is before the frame to be coded, The frame may be restricted not to refer to the previous frame in time. As a result, the in-plane predicted frame can be used as a random access start frame. Since the frames before the intra-predicted frame are not referred to, the frames can be correctly reproduced by decoding and reproducing the frames after the display time of the intra-predicted frame from the intra-predicted frame.

As described above, according to this embodiment, a more optimal prediction method can be selected by combining forward interpolation prediction and bidirectional prediction, so that the coding efficiency can be improved. Note that the first reference frame selection method, the second reference frame selection method, and the third reference frame selection method are not limited to one in advance, but a plurality of selection methods among the three methods are available. Block prediction type, if feasible
You may switch the several reference frame selection method by PredType.

(Second Embodiment) FIG. 8 is a block diagram of an image coding apparatus according to the second embodiment. Units that perform the same operations as those in the block diagram of the image coding apparatus according to the first embodiment in FIG. 1 and signals of the same operations are denoted by the same symbols, and description thereof will be omitted.

Pol is an interpolation prediction pixel interpolation means Pol for performing pixel interpolation. Since the prediction is switched on a block-by-block basis, forward interpolation prediction and bidirectional prediction of the image encoding device are not executed at the same time. Therefore, the means for performing the interpolation process can be shared by the forward interpolation prediction block and the bidirectional prediction block. Interpolation prediction pixel interpolation means Pol is
It is used for prediction image generation in both cases of the forward interpolation prediction block and the bidirectional prediction block. The switch SW1 is switched to the 0 side in the case of the forward interpolation prediction block and the bidirectional prediction block, and the output of the pixel interpolation means for interpolation prediction Pol is used as the prediction image signal Pred. The output of the multi-frame buffer MFrmBuf is switched to the 1 side and used as the predicted image signal Pred.

As described above, according to the present embodiment, the pixel interpolation means can be shared by the forward interpolation prediction block and the bidirectional prediction block, so that the structure of the image coding apparatus can be simplified.

(Third Embodiment) FIG. 9 is a block diagram of an image decoding apparatus according to the third embodiment. In FIG. 9, FIG.
2 is a block diagram showing a configuration of a conventional image decoding device, and units and signals having the same operation as those in the block diagram of the image encoding device according to the first embodiment shown in FIG. Is omitted. The image decoding device according to the present embodiment can be used, for example, for decoding a coded signal created by the image coding device according to any of the first and second embodiments.

The block prediction type PredType indicates a prediction type including in-plane prediction, forward prediction, forward interpolation prediction, backward prediction, and bidirectional prediction for a block. The multi-frame buffer MFrmBuf is 2 if the block prediction type PredType indicates forward interpolation prediction and bidirectional interpolation prediction.
Outputs the image signal of the block at the position indicated by each motion vector for one reference frame for generation of the predicted image signal Pred, and predicts the image signal of the block indicated by the motion vector for one reference frame if other Output for image signal Pred. If the block prediction type PredType indicates forward interpolation prediction, the switch SW is switched to the 0 side, the pixel value is interpolated by the forward prediction pixel interpolating means FwPol, and the predicted image signal Pred is output. Block prediction type Pr
When edType indicates bidirectional prediction, the switch SW is switched to the 2 side, and the bidirectional prediction pixel interpolating means BiPol generates the predicted image signal Pred. Block prediction type PredType
Shows other than forward interpolation prediction and bidirectional prediction, switch SW to 1 side and set the multi-frame buffer MFrmB.
The predicted image signal Pred of uf is used as it is. BiPol is a multi-frame buffer MFrmB
Pixel interpolation is performed by using two frames whose display times are before and after the frame to be coded in uf as reference frames and outputs a predicted image image Pred. The frame-type forward interpolation prediction reference frame selection unit FwPolRefFrmSel is a unit that selects a reference frame at the time of forward interpolation prediction for each type of frame, that is, a forward prediction frame and a bidirectional prediction frame. The method of selecting the reference frame is the same as that of the image coding apparatus according to the first embodiment. Reference frame selection method for forward interpolation prediction by frame type PolRefFrmS
el is a reference frame identifier at the time of forward interpolation prediction that indicates two reference frames used in the forward interpolation prediction block FwPolRef
The FrmID is notified to the reference frame selection means RefFrmSel. In the case of the forward interpolation prediction block, the reference frame selection unit RefFrmSel is used for the forward interpolation prediction reference frame identifier FwPo.
Using the frame indicated by lRefFrmID as a reference frame, the reference frame identifier RefFrmID is used to notify the reference frame to the multi-frame buffer MFrmBuf.

In the inter-frame prediction including the forward prediction frame and the forward interpolation prediction of the bidirectional prediction frame, if there is an in-plane prediction frame whose display time is before the encoding target frame, the in-plane prediction frame The frame may be restricted not to refer to the previous frame in time. As a result, the in-plane predicted frame can be used as a random access start frame. Since the frames before the intra-predicted frame are not referred to, the frames can be correctly reproduced by decoding and reproducing the frames after the display time of the intra-predicted frame from the intra-predicted frame.

As described above, according to the present embodiment, the image coded signal BitStrm output by the image coding apparatus using the image coding method described in the first and second embodiments can be correctly decoded.

It should be noted that, like the first embodiment, the first reference frame selection method, the second reference frame selection method, and the third reference frame selection method are not limited to one in advance. If a plurality of selection methods among the three methods can be realized, the plurality of reference frame selection methods may be switched by the block prediction type PredType.

(Fourth Embodiment) FIG. 10 shows a fourth embodiment.
3 is a block diagram of the image decoding device of FIG. Units and signals of the same operation as those in the block diagram of the image decoding apparatus according to the third embodiment in FIG. 9 are denoted by the same symbols, and the description thereof will be omitted. The image decoding device according to the present embodiment is, for example,
It can be used for decoding a coded signal created by the image coding apparatus according to the first or second embodiment.

Pol is an interpolation prediction pixel interpolation means Pol for performing pixel interpolation. Since the prediction is switched in block units, forward interpolation prediction and bidirectional prediction of the image decoding device are not executed at the same time. Therefore, the means for performing the interpolation process can be shared by the forward interpolation prediction block and the bidirectional prediction block. Interpolation prediction pixel interpolation means Pol is
It is used for prediction image generation in both cases of the forward interpolation prediction block and the bidirectional prediction block. The switch SW1 is switched to the 0 side in the case of the forward interpolation prediction block and the bidirectional prediction block, and the output of the pixel interpolation means for interpolation prediction Pol is used as the prediction image signal Pred. The output of the multi-frame buffer MFrmBuf is switched to the 1 side and used as the predicted image signal Pred.

As described above, according to the present embodiment, the pixel interpolation means can be shared by the forward interpolation prediction block and the bidirectional prediction block, so that the structure of the image decoding apparatus can be simplified.

(Fifth Embodiment) FIG. 11 shows a fifth embodiment.
3 is a block diagram of the image encoding device of FIG. Units that perform the same operations as those in the block diagram of the image coding apparatus according to the first embodiment in FIG.
The description is omitted.

A feature of the image coding method of the present invention is that a reference frame used for forward interpolation prediction is arbitrarily selected from a frame whose display time is before the frame to be coded in the multi-frame buffer, and the selected frame is selected. It is to store the identifier shown in the image coded signal.

The motion estimating means ME2, at the time of forward interpolation prediction,
2 reference frames and motion vectors for reference frames
Detect a pair. The motion vector obtained from the detection result is used as the motion information MV as the first reference frame number RefFrmNo1 indicating the first frame and the second reference frame number RefFrmNo2 indicating the second reference frame, and the block prediction type PredType
Output with. As the reference frame number, for example, a value numbered before the display time of the encoding target frame and in the closest order can be used. Variable length coding means
VLC2 is the coded residual signal ERes, reference frame number RefFrm
No1, RefFrmNo2, motion information MV, block prediction type PredTyp
e is variable-length coded and output as a coded image signal BitStrm2.

FIG. 12 is a conceptual diagram of frame selection during forward interpolation prediction according to the fifth embodiment. Frm1 to Frm7 indicate frames. Frm7 is the encoding target frame. CodingBlk
1 and CodingBlk2 indicate coding target blocks at different positions. A predictive image of CodingBlk1 is generated by forward interpolation prediction from blocks PredBlk11 and PredBlk12 that use two frames Frm3 and Frm4 as reference frames.
On the other hand, in CodingBlk2, a predicted image of CodingBlk2 is generated by forward interpolation prediction from blocks PredBlk21 and PredBlk22 using two frames Frm5 and Frm6 as reference frames. As shown in FIG. 10, different reference frames can be selected for blocks at different positions.

FIG. 13 is a conceptual diagram of the format of a coded image signal according to the fifth embodiment. The format of FIG. 13 shows the format of the forward interpolation prediction block. Frame is a signal for one frame, and Block is a signal for one block. PredType is a variable length coded signal of the block prediction type PredType. RefFrmNo1 and RefFrmNo2 are variable length coded signals of reference frame numbers for the first and second reference frames, respectively. Since the reference frame can be specified by the image coded signal BitStrm2 of this format, the image decoding device can correctly perform decoding. In the present embodiment, the first reference frame number Ref
FrmNo1, the second reference frame number RefFrmNop2 is variable length coded as an independent code in the image coded signal, but is coded as a multidimensional variable length code word in combination with other information in the image coded signal. May be.

As described above, according to the present embodiment, two reference frames for forward interpolation prediction can be selected from arbitrary frames, and there is a high possibility that a more optimal reference frame can be selected. Efficiency can be improved.

(Sixth Embodiment) In the image coding apparatus according to the fifth embodiment, one of the two reference frames is made common to all the forward interpolation prediction blocks of the frame to be coded,
Another sheet can be arbitrarily selected.
This can be considered as the case where the second reference frame number RefFrmNo2 does not exist in the block diagram of the image coding apparatus according to the fifth embodiment of FIG. The operation of the image decoding apparatus according to the sixth embodiment will be described with reference to the block diagram of FIG. At the time of forward interpolation prediction, the motion estimation means ME2 detects a motion vector for a reference frame common to all the forward interpolation prediction blocks of the encoding target frame and a motion vector for an arbitrarily selected reference frame. Further, the motion estimating means ME2 outputs the first reference frame number RefFrmNo1 indicating the selected one reference frame. The motion vector obtained from the detection result is output as motion information MV together with the first reference frame number RefFrmNo1 and the block prediction type PredType. The variable length coding means VLC2 is a coding residual signal ER.
es, reference frame number RefFrmNo1, motion information MV, block prediction type PredType is variable length coded, and image coded signal Bi
Output as tStrm2.

FIG. 14 is a conceptual diagram of frame selection during forward interpolation prediction according to the sixth embodiment. Frm1 to Frm7 indicate frames. Frm7 is the encoding target frame. CodingBlk
1 and CodingBlk2 indicate coding target blocks at different positions. Frm6 is a reference frame common to all the forward interpolation prediction blocks. Blocks PredBlk11 and Pre that use two frames Frm4 and Frm6 as reference frames
The predicted image of CodingBlk1 is generated by the forward interpolation prediction from dBlk12. On the other hand, in CodingBlk2, Frm5, Frm6
PredBlk that uses the two frames of
CodingBlk2 with forward interpolation prediction from 21 and PredBlk22
The predicted image of is generated. The common reference frame may be a frame whose display time is earlier than the encoding target frame and is closest to the encoding target frame in the multi-frame buffer.

FIG. 15 is a conceptual diagram of the format of a coded image signal according to the sixth embodiment. The format of FIG. 15 indicates the format of the forward interpolation prediction block. Frame is a 1-frame signal, Block is a 1-block signal, and Block prediction type PredType
And a reference frame identifier RefFrmID1 for a non-fixed reference frame is stored. Since the reference frame can be specified by the image coded signal BitStrm2 of this format,
The image decoding device can correctly perform decoding. In this embodiment, the first reference frame identifier RefFrmI
Although D1 is variable length coded as an independent code in the image coded signal, it may be coded as a multidimensional variable length code word in combination with other information in the image coded signal.

As described above, according to the present embodiment, one reference frame for forward interpolation prediction can be selected from arbitrary frames, and there is a high possibility that a more optimal reference frame can be selected. Efficiency can be improved.

(Seventh Embodiment) FIG. 16 shows a seventh embodiment.
3 is a block diagram of the image decoding device of FIG. In FIG.
Units that perform the same operations and signals that have the same operations as those in the block diagram of the image encoding device according to the second embodiment of FIG. 8 and the block diagram of the image decoding device according to the fourth embodiment of FIG. The description is omitted. The image decoding device according to the present embodiment can be used, for example, for decoding the coded signal created in the fifth embodiment.

The variable-length decoding means VLD2 performs variable-length decoding on the input coded image signal BitStrm2 to obtain a coded residual signal ERes,
1st reference frame number RefFrmNo1, 2nd reference frame number
It outputs RefFrmNo2, motion information MV, and block prediction type PredType. The reference frame selection means SRefFrmSel2 has a first reference frame number RefFrmNo1 and a second reference frame number RefFrmN.
By inputting o2, the frame indicated by the first reference frame number RefFrmNo1 and the second reference frame number RefFrmNo2 is designated as a reference frame in the multi-frame buffer MFrmBuf by the reference frame identifier RefFrmID. The forward interpolation prediction pixel interpolation means FwPol generates a predicted image Pred from the two designated reference frames.

As described above, according to the present embodiment, the image coded signal BitStrm2 output by the image coding apparatus using the image coding method described in the fifth embodiment can be correctly decoded.

(Embodiment 8) In the image decoding apparatus of Embodiment 7, one of the two reference frames is made common to all the forward interpolation prediction blocks of the frame to be coded, and the other one is arbitrarily set. It can also be made selectable. This can be considered to be the case where RefFrmNo2 does not exist in the block diagram of the image decoding apparatus according to the seventh embodiment in FIG. The eighth embodiment using the block diagram of FIG.
The operation of the image decoding apparatus will be described. The reference frame commonly used by all the forward interpolation prediction blocks is selected by the same method as that of the image coding device. The variable length decoding unit VLD2 performs variable length decoding on the input image coded signal BitStrm2, and outputs the coded residual signal ERes, the first reference frame number RefFrmNo1, the motion information MV, and the block prediction type PredType. The reference frame selection means SRefFrmSel2 is the first reference frame number RefFrm.
No1 is input and the frame indicated by the first reference frame number RefFrmNo1 and the frame commonly used by all the forward interpolation prediction blocks is the reference frame Reference frame identifier RefFrmID
Specify to the multi-frame buffer MFrmBuf by. The forward interpolation prediction pixel interpolation means FwPol generates a predicted image Pred from the two designated reference frames.

As described above, according to this embodiment, the image coded signal BitStrm2 output by the image coding apparatus using the image coding method described in the sixth embodiment can be correctly decoded.

(Ninth Embodiment) Furthermore, a program for realizing the configuration of the image coding method or the image decoding method shown in each of the above embodiments is recorded in a storage medium such as a flexible disk. As a result, the processing shown in each of the above embodiments can be easily implemented in an independent computer system.

FIG. 17 is an explanatory diagram of a storage medium for storing a program for realizing the image coding method or the image decoding method of the first to eighth embodiments by a computer system.

FIG. 17B shows the appearance, sectional structure, and flexible disk of the flexible disk as viewed from the front, and FIG. 17A shows an example of the physical format of the flexible disk, which is the recording medium body. .
The flexible disk FD is built in a case F, and a plurality of tracks Tr are formed concentrically from the outer circumference to the inner circumference on the surface of the flexible disk FD, and each track is angularly divided into 16 sectors Se. ing. Therefore, in the flexible disk storing the program, the image encoding method or the image decoding method as the program is recorded in the area allocated on the flexible disk FD.

FIG. 17C shows a structure for recording / reproducing the program on / from the flexible disk FD. When the program is recorded on the flexible disk FD, the image encoding method or the image decoding method as the program is written from the computer system Cs via the flexible disk drive FDD. Further, when the image coding method is constructed in the computer system by the program in the flexible disk, the program is read from the flexible disk by the flexible disk drive and transferred to the computer system.

In the above description, a flexible disk is used as a recording medium, but an optical disk may be used as well. Further, the recording medium is not limited to this, and any recording medium such as an IC card and a ROM cassette that can record the program can be similarly used.

[0070]

As described above in detail, according to the image coding method and the image decoding method of the present invention, forward interpolation prediction and bidirectional prediction can be combined and coded. As a result, a more optimal prediction method can be selected in inter-frame prediction, which is highly effective in improving coding efficiency.

[Brief description of drawings]

FIG. 1 is a block diagram of an image coding apparatus according to a first embodiment.

FIG. 2 is a conceptual diagram of frame selection during forward interpolation prediction in the forward prediction frame according to the first embodiment.

FIG. 3 is a conceptual diagram showing a multi-frame buffer at the time of forward interpolation prediction in the forward prediction frame according to the first embodiment.

FIG. 4 is a conceptual diagram of a first frame selection method at the time of forward interpolation prediction in a bidirectional prediction frame according to the first embodiment.

FIG. 5 is a conceptual diagram of a second frame selection method at the time of forward interpolation prediction in a bidirectional prediction frame according to the first embodiment.

FIG. 6 is a conceptual diagram of a third frame selection method at the time of forward interpolation prediction in a bidirectional prediction frame according to the first embodiment.

FIG. 7 is a conceptual diagram showing a multi-frame buffer at the time of forward interpolation prediction in a bidirectional prediction frame according to the first embodiment.

FIG. 8 is a block diagram of an image coding apparatus according to a second embodiment.

FIG. 9 is a block diagram of an image decoding device according to a third embodiment.

FIG. 10 is a block diagram of an image decoding device according to a fourth embodiment.

FIG. 11 is a block diagram of an image coding apparatus according to a fifth embodiment.

FIG. 12 is a conceptual diagram of frame selection during forward interpolation prediction according to the fifth embodiment.

FIG. 13 is a conceptual diagram of a format of a coded image signal according to the fifth embodiment.

FIG. 14 is a conceptual diagram of frame selection during forward interpolation prediction according to the sixth embodiment.

FIG. 15 is a conceptual diagram of a format of a coded image signal according to the sixth embodiment.

FIG. 16 is a block diagram of an image decoding device according to a seventh embodiment.

FIG. 17 is an explanatory diagram of a storage medium for storing a program for realizing the image encoding method or the image decoding method of the first to eighth embodiments by a computer system.

FIG. 18 is a block diagram showing the configuration of a conventional image encoding device.

FIG. 19 is a conceptual diagram of frame selection during forward interpolation prediction in the conventional image coding method.

FIG. 20 is an explanatory diagram of pixel value changes due to fade.

FIG. 21 is a conceptual diagram of a format of a coded image signal of a conventional image decoding device.

FIG. 22 is a block diagram showing the configuration of a conventional image decoding device.

[Explanation of symbols]

Img Image signal ImgEnc Image coding means ImgDec Image decoding means DImg Decoded image Add Adder Sub Subtractor MFrmBuf Multi-frame buffer ME, ME0, ME1 Motion estimation means VLC, VLC0, VLC1 Variable length coding means VLD, VLD0, VLD1 Variable length Decoding means RefFrmSel Reference frame selection means FwPol Forward interpolation prediction pixel interpolation means BiPol Bidirectional prediction pixel interpolation means Pol Interpolation prediction pixel interpolation means RefFrmSel Reference frame selection means FwPolRefFrmSel Frame type forward interpolation prediction reference frame selection means Cs Computer System FD Flexible Disk FDD Flexible Disk Drive

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Toshishi Kondo             1006 Kadoma, Kadoma-shi, Osaka Matsushita Electric             Sangyo Co., Ltd. F-term (reference) 5C059 LB16 LB18 MA00 MA04 MA05                       MA14 MA23 ME01 PP05 PP06                       PP07 UA02 UA05 UA33                 5J064 AA01 BA09 BB04 BC01 BC08                       BC14 BC29 BD03

Claims (20)

[Claims] 1. A bidirectional prediction frame that performs inter-frame prediction from frames before and after a display time, a forward prediction frame that performs inter-frame prediction from only a frame at a previous display time, and an inter-frame prediction as prediction frames. In an image coding method including an in-plane prediction frame that does not perform, when the coding target frame is a forward prediction frame, or a bidirectional prediction frame, two frames whose display time is earlier than the coding target frame An image coding method characterized in that a predicted pixel value is generated by pixel interpolation using as a reference frame. 2. The encoding target frame is a forward prediction frame, and the display time is earlier than the encoding target frame.
2. The image coding method according to claim 1, wherein, when inter-frame prediction is performed by pixel interpolation using one frame as a reference frame, each of the two reference frames is a forward prediction frame or an in-plane prediction frame. 3. The encoding target frame is a bidirectional prediction frame, and the display time is earlier than the encoding target frame.
The image coding according to claim 1, wherein when the two frames are used as reference frames for interframe predictive coding by pixel interpolation, each of the two reference frames is a forward prediction frame or an in-plane prediction frame. Method. 4. The encoding target frame is a bidirectional prediction frame, and the display time is before the encoding target frame.
2. The image code according to claim 1, wherein when performing inter-frame prediction by pixel interpolation using two frames as reference frames, at least one of the two reference frames is a forward prediction frame or an in-plane prediction frame. Method. 5. The encoding target frame is a bidirectional prediction frame, and the display time is earlier than the encoding target frame.
When performing inter-frame prediction by pixel interpolation using two frames as reference frames, the two reference frames are frames in the multi-frame buffer storing encoded frames before the display time of the encoding target frame, and The image encoding method according to claim 1, wherein the two images are displayed in the order of closest display time. 6. A predictive frame, a bidirectional predictive frame for performing inter-frame predictive from frames before and after display time, a forward predictive frame for performing inter-frame predictive only for a frame at previous display time, and an inter-frame predictive frame. In the image decoding method including the in-plane prediction frame that does not perform, when the frame to be decoded is a forward prediction frame or a bidirectional prediction frame, the two frames whose display time is earlier than the encoding target frame are An image decoding method characterized by generating a predicted pixel value by pixel interpolation as a reference frame. 7. When the frame to be decoded is a forward prediction frame and two frames whose display time is earlier than the frame to be decoded are used as reference frames to perform inter-frame prediction by pixel interpolation, each of the two reference frames Is a forward prediction frame or an in-plane prediction frame. 8. When the frame to be decoded is a bidirectional prediction frame and two frames whose display time is earlier than the frame to be decoded are used as reference frames to perform inter-frame prediction by pixel interpolation, each of the two reference frames The image decoding method according to claim 6, wherein the frame is a forward prediction frame or an in-plane prediction frame. 9. The two reference frames when the decoding target frame is a bidirectional prediction frame and inter-frame prediction is performed by pixel interpolation using two frames whose display time is earlier than the decoding target frame as reference frames. 7. The image decoding method according to claim 6, wherein at least one of the frames is a forward prediction frame or an in-plane prediction frame. 10. The decoding target frame is a bidirectional prediction frame, and the display time is earlier than the decoding target frame.
When inter-frame prediction is performed by pixel interpolation using one frame as a reference frame, the two reference frames are the frames in the frame buffer storing the decoded frames, and the order closest to the display time of the encoding target frame. 7. The image decoding method according to claim 6, wherein the number of images is 2. 11. An image coding method for predicting and coding a moving image by inter-frame prediction, in the case of inter-frame prediction by pixel interpolation using two frames whose display time is earlier than a coding target frame as reference frames. , An image code characterized by arbitrarily selecting two reference frames from a frame whose display time is earlier than the encoding target frame and storing information for identifying the two reference frames in an image encoded signal Method. 12. An image coding method for predicting and coding a moving image by inter-frame prediction, in the case of inter-frame prediction by pixel interpolation using two frames whose display time is earlier than a frame to be coded as reference frames. , A frame determined by a predetermined method and a frame whose display time is earlier than the frame to be encoded are arbitrarily selected as a reference frame, and the frame is arbitrarily selected from the two reference frames. An image coding method characterized in that only information for identifying a reference frame is stored in an image coded signal. 13. An image decoding method for predicting and decoding a moving image by inter-frame prediction, input in the case of inter-frame prediction by pixel interpolation using two frames whose display time is earlier than the decoding target frame as reference frames. An image decoding method characterized by decoding identification information for identifying two reference frames from an encoded image signal and using the two frames indicated by the identification information as reference frames for prediction by pixel interpolation. . 14. An image decoding method for predicting and decoding a moving image by inter-frame prediction, input in the case of inter-frame prediction by pixel interpolation using two frames whose display time is earlier than the decoding target frame as reference frames. The identification information for identifying one reference frame is decoded from the image coded signal, and pixel interpolation is performed using two frames, the frame indicated by the identification information and the reference frame determined by a predetermined method, as reference frames. An image coding method characterized by being used for prediction. 15. An image coding apparatus for performing inter-frame prediction, comprising: a multi-frame buffer that stores coded frames; and two frames whose display time is earlier than a frame to be coded in the multi-frame buffer. Forward prediction pixel interpolating means for pixel interpolation as a reference frame, bidirectional prediction pixel interpolating means for pixel interpolating two frames whose display time before and after the encoding target frame is a reference frame in the multi-frame buffer, Forward frame reference frame selection means for selecting a reference frame for forward interpolation prediction by a predetermined method, and forward frame prediction reference for each frame type when inter-frame prediction is performed by the forward prediction pixel interpolation means. Equipped with motion estimation means for performing motion estimation using the reference frame selected by the frame selection means The image coding apparatus, characterized in that the at it. 16. An image decoding apparatus for performing inter-frame prediction, wherein a multi-frame buffer for accumulating decoded frames and two frames whose display time is earlier than the decoding target frame in the multi-frame buffer are used as reference frames. Pixel interpolating unit for forward prediction for pixel interpolation, Pixel interpolating unit for bidirectional prediction for interpolating pixels using the two frames whose display times before and after the decoding target frame in the multi-frame buffer are reference frames, and for forward interpolation prediction Frame type-based forward prediction reference frame selection means for selecting the reference frame by a predetermined method, and the frame type-based forward interpolation prediction reference frame selection means for inter-frame prediction by the forward prediction pixel interpolation means. It is equipped with a reference frame selection means that uses the selected reference frame. Image decoding apparatus according to claim. 17. A storage medium that stores a program for performing image encoding by a computer, the program being a bidirectional prediction for performing inter-frame prediction from a frame whose display time is before and after as a prediction frame to the computer. Frames and forward-predicted frames that perform inter-frame prediction only from the frame with the previous display time, and
In an image coding method including an intra-frame prediction frame that does not perform inter-frame prediction, when the coding target frame is a forward prediction frame or a bidirectional prediction frame,
A storage medium characterized in that a predicted pixel value is generated by pixel interpolation using two frames whose display time is earlier than the encoding target frame as reference frames. 18. A storage medium storing a program for performing image decoding by a computer, wherein the program causes the computer to perform a bidirectional prediction frame for performing inter-frame prediction from a frame before and after a display time as a prediction frame. , And, in the image decoding method including the forward prediction frame that performs inter-frame prediction only from the frame whose display time is previous, and the in-plane prediction frame that does not perform the inter-frame prediction, the decoding target frame is the forward prediction frame, or When the frame is a bidirectional prediction frame, it is characterized in that a predicted pixel value is generated by pixel interpolation using two frames whose display time is before the encoding target frame as reference frames. Storage medium. 19. A storage medium storing a program for performing image encoding by a computer, wherein the program is encoded in a computer by an image encoding method for predicting and encoding a moving image between frames. In the case of inter-frame prediction by pixel interpolation that uses two frames whose display time is earlier than the target frame as reference frames, the frame whose display time is earlier than the encoding target frame is arbitrarily
A storage medium for selecting two reference frames and storing information for identifying the two reference frames in an image coded signal. 20. A storage medium storing a program for performing image decoding by a computer, wherein the program is displayed on a computer from a decoding target frame in an image decoding method of predicting and decoding a moving image between frames. In the case of inter-frame prediction by pixel interpolation using the two previous frames of time as reference frames, the identification information for identifying the two reference frames from the input image coded signal is decoded, The storage medium characterized in that the two frames shown by are used for prediction by pixel interpolation as reference frames.