JPH11289541A - Moving image coding method, moving image coder and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reproduce a program before a skip start point and a program after a skip arrival point seamlessly in the case of skipping part of a moving image program to designate the reproduction. SOLUTION: An analysis section calculates a bit occupancy amount boe of a virtual decoder after the end of decoding a final picture of GOP-new-0 under the assumption of encoding the GOP-new-0 with a decided assignment bit amount (step S24). Then the size of boe is compared with that of bis (step S25) and when it is discriminated that the boe is over the bis, a B0 of GOP-0 is re-encoded to an I0x of an I picture and a B1 of an out-point-picture is re- encoded to a P1x of the P picture, a new GOP-new-0 is generated (step S26).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a moving picture coding method, a moving picture coding apparatus, and a recording medium for processing digital signals compressed and coded by using orthogonal transform such as discrete cosine transform.

[0002]

2. Description of the Related Art In recent years, motion compensation (MC) processing, discrete cosine transform (de
MPEG (moving p) combining redundancy reduction processing by orthogonal transform such as discrete cosine transformation (DCT)
and the MPEG2 have become widely used.

[0003] In MPEG2, in order to perform predictive coding using motion compensation, a GOP (group of pictures) composed of three elements called an I picture, a P picture, and a B picture.
Structure is used. I picture (intra coded image)
Are generated by intra-picture coding, and do not perform predictive coding from the previous picture.

[0004] If only images created using predictive coding are arranged, if a random access is made, a screen cannot be instantaneously displayed according to the random access. To support random access. That is, the I picture is G
This is to keep the OP independent.

[0005] The appearance frequency of I-pictures is determined by the performance of random access required for an application, but is usually one per GOP, for example, 2 GO per second.
In the case of P, the ratio is one for every 15 frame images. A GOP refers to a group of pictures from one I picture to the next I picture. Image prediction is performed between pictures in this group.

[0006] The P picture is (predictive coded image)
Is made with predictive coding from the previous image, and is made based on I-pictures. A P picture is defined as an "inter-frame forward prediction coded image" with respect to an I picture which is an "intra-frame coded image". The B picture (bidirecitionlly predictive coded image) is a “bidirectional predictive coded image”. A B picture is created by performing prediction from two preceding and succeeding P pictures.

G of I picture, P picture and B picture
Looking more closely at the relationship within the OP, the first-stage prediction from the I-picture at the beginning of the GOP is made forward, that is, in the forward direction, to produce a P-picture. At this time, P
The pictures are arranged so as to skip over a B picture created later.

The second-stage prediction is performed between the first P-picture and the P-picture coded in the first stage and between the two, that is, one or a plurality of B-pictures depending on bidirectional prediction. A picture is created. Further, a B picture is similarly created between the first P picture and the second P picture. The B picture has 2 bits for decoding.
Motion compensation is performed using the motion vector of the book and the two preceding and succeeding reference images.

In bidirectional prediction, which is a feature of MPEG, a high prediction efficiency is obtained because two images, a temporally past image and a future image, are used in the prediction.

[0010]

By the way, MPEG2
In some cases, it is specified that a part of the moving image program is skipped and reproduced. The reproduction of the MPEG2 image is performed by the above-described MPEG2 encoding method.
Generally, this is performed by predicting and referring to the pictures before and after the image.

For this reason, in general, it is necessary to predict and refer to a picture behind the skip start point and a picture before the skip arrival point. Therefore, it is necessary to seamlessly reproduce the skip arrival point from the skip start point seamlessly. Was difficult.

For example, when the in-point picture is a B picture, the I picture necessary for decoding the B picture is
It is necessary to decode the picture or the P picture, and in this case, there is a possibility that the reproduced image is interrupted.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-described problems, and has been made in consideration of a situation in which a reproduction of a part of a moving image program is designated to be skipped. A moving picture coding method, a moving picture coding apparatus, and a recording method capable of efficiently re-encoding a GOP before a skip start point and a GOP after a skip arrival point in order to seamlessly reproduce a program on the rear side. The purpose is to provide a medium.

[0014]

According to the moving picture encoding method of the present invention, an input bit stream is an image of an image group including an image at an out point, and is output from the first display image of the image group. Each image up to the point is re-encoded into an inter-picture prediction coded image predicted from the image of the image group displayed immediately before the image group including the image at the out point, and The image processing apparatus further includes a reconstruction step of reconstructing a display image into a group of images that can be decoded without predictive reference.

According to the moving picture coding apparatus of the present invention, for an input bit stream, each of the images from the first display image to the out point of the image group including the out-point image is converted. Re-encode to an inter-picture prediction coded image predicted from the image of the image group displayed immediately before the image group including the image at the out point, and predictively refer to the display image after the image at the out point. A reconstructing means for reconstructing a group of images that can be decoded without decoding.

According to the recording medium of the present invention, for an input bit stream, each of the images from the first display image to the out point of the image group including the out-point image is converted to the out-point image. Re-encode to an inter-picture prediction coded image predicted from the image of the image group displayed immediately before the image group including the image of, and decode the display image subsequent to the out-point image without predictive reference A control program for reconstructing an image group that can be reconstructed is recorded.

In the moving picture coding method according to the present invention, for an input bit stream, the first intra-coded picture of the picture group displayed one behind the picture group including the picture at the in-point is converted to Re-encoding to an inter-picture prediction coded image predicted from the images of the image group including the point image, and reconstructing the display image preceding the in-point image into an image group that can be decoded without predictive reference It has a configuration step.

According to the moving picture coding apparatus of the present invention, for an input bit stream, the first intra-coded picture of the picture group displayed one behind the picture group including the picture at the in-point is converted to the Re-encoding to an inter-picture prediction coded image predicted from the images of the image group including the point image, and reconstructing the display image preceding the in-point image into an image group that can be decoded without predictive reference It is characterized by having constitution means.

According to the recording medium of the present invention, for an input bit stream, the first intra-coded image of an image group displayed after the image group including the image at the in-point is converted to the image at the in-point. A control program that re-encodes the inter-prediction prediction image predicted from the image of the image group including the image and reconstructs the display image preceding the in-point image into a group of images that can be decoded without predictive reference is recorded. It is characterized by having been done.

According to the moving picture coding method of the present invention, in an input bit stream, an image group up to a first point in a unit section including a first point in accordance with a display order and display are performed. The image group of the previous unit section according to the order is
An image behind the first point is reconstructed into a first image group that can be decoded without predictive reference, and a second image group in a unit section including the second point according to the display order. Reconstruction for reconstructing a group of images from a point and a group of images in the next unit section in accordance with the display order into a second group of images that can be decoded without predicting and referring to the image before the second point It is characterized by having a process.

According to the moving picture coding apparatus of the present invention, in the input bit stream, a group of images up to a first point in a display order in a unit section including the first point in accordance with a display order are displayed. The image group of the previous unit section according to the order is
An image behind the first point is reconstructed into a first image group that can be decoded without predictive reference, and a second image group in a unit section including the second point according to the display order. Reconstruction for reconstructing a group of images from a point and a group of images in the next unit section in accordance with the display order into a second group of images that can be decoded without predicting and referring to the image before the second point It is characterized by having means.

According to the recording medium of the present invention, the input bit stream follows the display order and the image group up to the first point in the unit group including the first point in the unit order. The image group of the previous unit section is reconstructed into a first image group that can be decoded without predicting and referring to the image behind the first point, and the image of the unit section including the second point Decoding the image group from the second point in the group according to the display order and the image group in the next unit section according to the display order without predicting and referring to the image before the second point. A control program for reconstructing a second image group to be obtained is recorded.

In the moving picture coding method according to the present invention, when the first point is a bidirectional predictive coded picture, a group of pictures in a unit section including the first point is decoded and the first group is decoded according to the display order. If the second point is a forward predictive coded image or a bidirectional predictive coded image, the image group of the unit section including the second point is decoded, and the second point is decoded according to the display order. And a reconstructing step of re-encoding the second and subsequent points.

When the first point is a bidirectionally predicted coded image, the moving image coding apparatus according to the present invention decodes a group of images in a unit section including the first point and follows the first group in accordance with the display order. If the second point is a forward predictive coded image or a bidirectional predictive coded image, the image group of the unit section including the second point is decoded, and the second point is decoded according to the display order. And reconstructing means for re-encoding the second and subsequent points.

In the recording medium according to the present invention, when the first point is a bidirectional predictive coded image, the image group of the unit section including the first point is decoded, and the first point according to the display order is decoded. If the second point is a forward predictive coded image or a bidirectional predictive coded image, the second point is decoded according to the display order by decoding the group of images in the unit section including the second point. A control program for re-encoding the data after the point is recorded.

[0026]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a moving picture coding method, a moving picture coding apparatus and a recording medium according to the present invention will be described in detail with reference to the drawings.

First, a first embodiment of the present invention will be described. The processing of the bit stream according to the invention comprises:
As shown in the coded bit stream of A in FIG. 1, when specifying to skip and reproduce a part of the moving image program, the time axis is shifted from the out point which is the first point which is the skip start point. The program on the front side and the program on the rear side of the in point, which is the second point, which is the skip arrival point, in the display order, which is the direction, can be reproduced seamlessly.

In this moving picture coding method, the moving picture program is given as a coded bit stream of the MPEG2 standard.

The MPEG2 standard coded bit stream has a unit of GOP (group of pictures), which is a group of pictures in which coded pictures are arranged in a predetermined order. That is, in A in FIG. 1, B04 which is an out-point picture is included in GOP-0 which is an out-point side GOP following GOP-(− 1), and Pn5 which is an in-point picture is an in-point side GOP. GOP- (n + 1) is included in a certain GOP-n, and this in-point side GOP is followed by GOP- (n + 1). Here, GOP-i is the i-th GOP in the time axis direction, which is the display order of images.

A GOP is obtained by encoding an image using an I (intra) picture, which is a reference image in which an image is encoded without predictive encoding from another image, and predictive encoding in the forward direction in the display order. P (predicti)
ve) B, which is a bidirectional prediction coded image in which a picture is coded using forward and backward prediction coding
(Bidirectionally) pictures.

The GOP is composed of an I picture, a plurality of P pictures, and a
Pictures are arranged in a predetermined order.

In the coded bit stream of A in FIG. 1, GOP-0 which is the out-point side GOP is I0
2, B00, B01, P05, B03, B04, P0
8, B06, and B07. GOP-n which is the in-point side GOP
Are In2, Bn0, Bn1, Pn5, Bn3, Bn
4, Pn8, Bn6, and Bn7.

Here, Iij is the I-th picture in the display order in the i-th GOP, and Pij is the i-th G picture.
The P-picture whose display order in the OP is j-th is Bij
Indicates the B-picture whose display order in the i-th GOP is j-th.

With respect to the coded bit stream as indicated by A in FIG. 1, an image on the front side in the display order from the out point and an image on the rear side from the in point can be reproduced seamlessly.

For this purpose, GOP-0, which is the out-point side GOP including B04, which is the out-point picture Pout, is decoded (decoded). Subsequently, the GOP-0 is decoded so that the out-point picture B04 is decoded without predicting and referring to the next coded image in the display order.
Is again encoded. That is, in the sequence of the encoded bit stream after the re-encoding shown in B in FIG. 1, the GO point on the out point side shown in A in FIG.
GOP-0 which is P is the re-encoded GOP-n
ew-0.

The GOP-n, which is the in-point side GOP including the in-point picture Pn5, is decoded (decoded).
I do. Subsequently, the in-point picture Pn5
Is encoded (encoded) again so that the GOP-n can be decoded without predicting and referring to the previous encoded image in the display order. That is, in B in FIG. 1, GOP-n which is the GOP on the out-point side shown in A in FIG. 1 corresponds to the re-encoded GOP-new-n.

As a result of the re-encoding of the GOP on the in-point side and the GOP on the out-point side, as shown in C in FIG. 1, the frames to be reproduced and F00, F01, F02, F
As in 03, F04, Fn5, Fn6, Fn7, and Fn8, a frame before the IN point and a frame after the OUT point are seamlessly reproduced seamlessly.

Here, GOP-new-i is GOP-new-i.
-A new GOP created by re-encoding some pictures of i.

For example, GOP-new-0 in this example is
Pictures B00, B01, I02, B0 of GOP-0
3, B04 decoded frames F00, F01, F0
This is a new GOP created from 2, F03 and F04.
GOP-new-n is Pn5, Bn6, of GOP-n.
Frames Fn5, Fn6, and Fn obtained by decoding Bn7 and Pn8
This is a new GOP made from n7 and Fn8. Here, the frame Fij corresponds to a picture having the same subscript.

The above-described series of procedures relating to the MPEG bit stream constitutes a reconstruction process.

Next, the re-encoding of the GOP on the out-point side will be specifically described.

When the out-point picture is an I-picture or a P-picture, the out-point side GOP does not necessarily need to be decoded and re-encoded. In this case, the picture following the P picture or the I picture following the out-point picture may be removed.

For example, in the sequence of the coded bit stream indicated by A in FIG.
If the picture P05 of OP-0 is an out-point picture, GOP-new-0 can be created by removing pictures after picture P08 as follows.

[0044]

(Equation 1)

When the out-point picture is a B picture, the out-point side GOP needs to be decoded and then re-encoded. The GOP is re-encoded so that the out-point picture can be decoded without predicting reference to a picture subsequent to the out-point picture in display order.

For example, as shown at A in FIG.
B04 of GOP-0 which is OP is out point picture Po
ut, the pictures B03 and B04 that have been predictively coded from the pictures I02 and P05 are
5 is re-encoded so that it can be created without predictive reference.

First, I02, B00, B01, P0
5, B03 and B04, and decodes frames F00 and F0.
1, F02, F04, F05, and then the frame F0
0, F01, F02, F03, and F04 are re-encoded.

As shown below, GOP-new-0
In (a), F03 and F04 are replaced with P03 and P of the P picture.
04 is re-encoded. In this case, P03 is I0
2, P04 is predictively coded from P03.

In GOP-new-0 (b), F
04 is re-encoded to P04 of a P picture, and F03 is B0 of a B picture to be predictively coded from I02 and P04.
3 is re-encoded.

Note that GOP-new-0 (a) and GOP
−new-0 (b), I02, B00, B02
Is preferably copied from GOP-0 without performing re-encoding.

[0051]

(Equation 2)

Here, the subscripts of the characters I, B and P correspond to the image F having the same subscripts.

Next, the re-encoding of the GOP on the in-point side will be specifically described.

If the in-point picture is an I-picture, the GOP on the in-point side does not necessarily need to be decoded and re-encoded.

In this case, pictures after the in-point picture may be copied and used as they are.

For example, the in-point side GOP indicated by A in FIG.
If the picture In2 of GOP-0 is an in-point picture, the pictures subsequent to the picture In2 are
OP-new-0.

When the in-point picture is a P picture or a B picture, the GOP on the in-point side needs to be re-encoded. The GOP is re-encoded so that the in-point picture can be decoded without predictive reference to a picture preceding the in-point picture in display order.

For example, when Pn5 is an in-point picture Pin like GOP-n of A in FIG. 1, re-encoding is performed so that Pn5 that has been predictively coded from In2 can be created without predicting and referring to In2. I do. at first,
GOP-n pictures In2, Bn0, Bn1, Pn
5, Bn3, Bn4, Pn8, Bn6, Bn7, and decode frames Fn0, Fn1, Fn2, Fn3, Fn.
4, Fn5, Fn6, Fn7, and Fn8, and then re-encode the frames Fn5, Fn6, Fn7, and Fn8.

As shown below, the subscripts of I, B, and P of GOP-new-n correspond to the image F having the same subscript. In GOP-new-n, Fn5 is re-encoded to In5 of an I picture.

[0060]

(Equation 3)

In another example, when Bn3 of GOP-n is an in-point picture, Fn5 is set to I-point as follows.
It is good to re-encode to In5 of the picture.

[0062]

(Equation 4)

That is, in GOP-n, the first P picture Pn5 after the picture of the in-point picture Bn3 in the display order is re-encoded into the I picture In5.

In GOP- (n + 1) following the re-encoded GOP-new-n, closed_go
When p = 0, the first B picture of GOP- (n + 1) predicts and references the original GOP-n picture.

Here, closed_gop is the GO of MPEG.
This is a 1-bit flag in the P header. closed_gop
Indicates the type of prediction used in the B picture immediately following the first I picture following the GOP header.

When closed_gop is set to “1”, GO
Indicates that the B picture immediately following the first I picture following the P header was coded using only backward prediction or intra (intra-picture) coding.

For example, in the following image sequence, Ix
When the closed_gop of the GOP starting from 2 is set to “1”, Bx0 and Bx1 are coded using only backward prediction from Ix2 or intra coding, and Pa3
This indicates that forward prediction from and bidirectional prediction from Pa3 and Ix2 are not used. Here, "|"
Indicates the boundary of. Ia0 Pa3 Ba1 Ba2 | Ix2 Bx0 Bx1
When Px5 Bx3 Bx4closed_gop is set to “0”, the GOP refers to the picture of the immediately preceding GOP.

GOP-n re-encoded from GOP-n
Since the image quality of ew-n is different from the original GOP-n, there is a possibility that the image quality of the first B picture of GOP- (n + 1) is strictly mismatched. However, this GO
The effect on the picture quality of the first B picture of P- (n + 1) is as follows.
It seems that there is almost no problem. Of course, if this effect is considered serious, the first BOP- (n + 1) of GOP- (n + 1) that predicts and references the original GOP-n picture
Pictures may also be re-encoded.

Next, restrictions on the re-encoding condition will be described. When re-encoding GOP-0 which is an out-point side GOP and GOP-n which is an in-point side GOP to obtain GOP-new-0 and GOP-new-n,
It is necessary to take care so that the VBV buffer, which is a virtual buffer of the decoder defined by MPEG, does not break down when continuous playback is performed from GOP-new-0 to GOP-new- (n + 1). The fact that the VBV buffer does not fail means that the buffer does not overflow or underflow.

If the application that reproduces the re-encoded bit stream can control the input rate to the VBV buffer and can prevent the buffer from overflowing, it is only necessary to prevent the buffer from underflowing. You have to be careful. In the following example, conditions for re-encoding so that the buffer does not underflow will be described.

In this case, the first I of GOP-new-n
The coded bit amount of a picture is most restricted. That is, a VB with a buffer size of 1.75 Mbit shown in FIG.
In the bit occupancy of the V buffer, the first I picture of GOP-new-n in which GOP-n, which is the in-point side GOP including the in-point picture, is re-encoded, must be smaller than the size indicated by D in the figure. Must.

The greater the input bit rate of the VBV buffer, the smaller the restrictions on re-encoding GOP-new-n. Generally, the maximum bit rate determined by the application is the input bit rate of the VBV buffer.

In the case of MPEG @ MP (main profile / main level) of MPEG2 corresponding to so-called DVD-Video, digital satellite broadcasting, etc., the maximum value of the input bit rate of the VBV buffer is 15 Mbps. When 15 Mbps can be secured when digitally outputting a bit stream to an IEEE 1394 standard digital bus by isochronous transfer, the input bit rate of the VBV buffer at the time of re-encoding is set to 15 bp.
It may be s. However, in practice, a bit rate that is the sum of the bit rates of audio and multiplexing overhead is required.

The minimum value of the bit amount of the first I picture of GOP-new-n, that is, the worst pattern is (input bit rate of VBV buffer / frame rate). In this case, it is difficult to encode GOP-new-n with high image quality. The situation close to the worst pattern is that the picture at the out point is an I picture
This may occur when the number of pictures of P-new-0 is as small as 2 to 3 pictures.

Next, a method for relaxing the above-described restriction on re-encoding will be described.

As shown by A in FIG. 3, the pictures in the hatched portions, that is, GOP-(-1) and GOP-0, which are the two GOPs before the edit out point, and the in-point The last two GOPs, GOP-n and GOP-
(N + 1) is re-encoded, and GOP-new-A and GOP-new-
In the case of B, the restrictions on the encoding of pictures behind the in-point picture are relatively relaxed.

In order to perform encoding with high image quality, an I picture requires a much larger amount of coding bits than a P picture or a B picture. However, when the interval between I-pictures is reduced to, for example, two to three pictures, in order to prevent buffer underflow of the VBV buffer, the coded bit amount of the I-picture has to be smaller than the bit amount necessary for high image quality. Disappears. In this case, GO
It is difficult to encode P.

In the out-point side program in the example shown in FIG. 1, when the out-point side picture is an I-picture,
When the number of pictures of P-new-0 is small, for example, 2 to 3 pictures, or when the number of pictures of GOP-new-n is small, for example, 2 to 3 pictures in the program on the in-point side, the worst pattern described above is obtained. A close situation can occur. In this case, GOP-new-
It is difficult to encode 0 or GOP-new-n with high image quality.

In such a case, as shown in FIG. 3, in the out-point side program, the pictures in the hatched portions, that is, the two GOPs before the edit out-point picture are converted into one GOP- If re-encoding as new-A, the image quality can be improved. Further, in the in-point side program, the picture in the hatched portion, that is, two GOPs after the in-point side picture are re-encoded as one GOP-new-B, so that the image quality can be improved. it can.

An example of improving the image quality of the re-encoding of the out-point side program will be described below. First, it is assumed that GOP-(-1) and GOP-0 are configured as follows.

I2 B0 B1 P5 B3 B4 P8 B6 B7 Pb B9 Ba |
I2 B0 B1 P5 B3 B4 P8 B6 B7 PbB9 Ba Note that the symbol “|” represents the boundary of GOP, the left side of “|” is GOP-(− 1), and the right side is GOP-.
0.

Next, the out-point picture is changed to the GOP-0
When GOP-0 is re-encoded as I2, GOP-new-0 is configured as follows.

I2 B0 B1 P5 B3 B4 P8 B6 B7 Pb B9 Ba | I2 B0 B1 GOP-new-0 is on the right side of the symbol “|”.

Then, in order to improve the image quality, two GOPs on the out point side, GOP-(-1) and GOP-ne, are used.
Re-encoding w-0 into one GOP-new-A is as follows.

I2 B0 B1 P5 B3 B4 P8 B6 B7 Pb B9 Ba P2 B0 B1 That is, GOP-0 including the above out-point picture I2
The first I picture I2 of (GOP-new-0) is re-encoded into a P picture P2 predicted from the GOP-(-1) picture immediately preceding the GOP-0.

An example of improving the image quality of the re-encoding of the in-point side program will be described below.

First, GOP-n and GOP- (n +
Assume that 1) is configured as follows.

I2 B0 B1 P5 B3 B4 P8 B6 B7 Pb B9 Ba |
I2 B0 B1 P5 B3 B4 P8 B6 B7 PbB9 Ba GOP-n is on the left side of the symbol “|”, and GOP- (n + 1) is on the right side.

Next, the in-point picture is changed to B9 of GOP-n.
When GOP-n is re-encoded, GOP-new-n is configured as follows.

Ib B9 Ba | I2 B0 B1 P5 B3 B4 P8 B6 B7 Pb B9 Ba GOP-new-n is on the left side of the symbol “|”.

Then, in order to improve the image quality, two points on the in-point side
GOPs, GOP-new-n and GOP- (n
+1) is re-encoded into one GOP-new-B as follows.

Ib B9 Ba P2 B0 B1 P5 B3 B4 P8 B6 B7 Pb B9 Ba That is, GOP-n (GOP) including in-point picture B9
-New-n) and GOP- (n +
GOP-n (GOP-n)
ew-n) is re-encoded into a P picture P2 predicted from the picture.

When the number of pictures in the GOP is limited by the application, GOP-new-A, GOP
The length of OP-new-B may exceed the limit. In this case, GOP-new-A, GOP-
New-B may be encoded by dividing it into two GOPs.

For example, in the out-point side program, before the improvement, I2 B0 B1 P5 B3 B4 P8 B6 B7 Pb B9 Ba | I2 B0 B1, but after the improvement, I2 B0 B1 P5 B3 B4 P8 B6 B7 | Ib B9 Ba P2 B0 B1. That is, two GOPs, GOP-(-
1) and GO in which GOP-new-0 is two GOPs
It is re-encoded separately into P-new-A1 and GOP-new-A2.

In the in-point side program, I2 B0 B1 | I2 B0 B1 P5 B3 B4 P8 B6 B7 Pb B9 Ba before the improvement, but I2 B0 B1 P2 B0 B1 | I5 B3 B4 P8 after the improvement B6 B7 Pb B9 Ba. That is, two GOPs, GOP-new
GO where -n and GOP- (n + 1) are two GOPs
It is re-encoded separately into P-new-B1 and GOP-new-B1.

These examples are based on the application
This is a case where the maximum value of the number of pictures of the OP is limited to twelve.

Next, the system (multiplexed) stream will be described.

In the MPEG system, data to be multiplexed is divided into a stream of an appropriate length called a packet, and the packet is switched by adding additional information such as a header and time-divisionally transferred.

In the MPEG synchronization system, information called a time stamp indicating when decoding / reproduction should be performed is added to each decoding / reproduction unit called an access unit. And for the time stamp, the SCR
(System Clock Reference)
Information provides a time reference.

The time stamp includes PTS (Presentati
There is reproduction management time information called reproduction on time stamp (DTS) and decoding management information called DTS (Decoding Time Stamp).

In the section to be re-encoded, the system encoding is also newly added, so that a new time stamp is added. However, in this case, the time stamp (SCR,
(PTS, DTS) discontinuity occurs.

To make the time stamps continuous, it is necessary to remultiplex the entire program and replace the time stamps. However, depending on the length of the program time,
This method is not practical for applications such as editing on storage media because of the length of time required for remultiplexing.

Therefore, as shown in FIG. 4, the re-multiplexing section is limited to the re-encoded section, and the new-stc
A new time stamp will be added only to the section. Here, the section between stc1 and stc2 is the same as the time stamp originally attached to the multiplexed stream.

In the new-stc section, system encoding is performed using an STC (System Time Clock) continuous with stc1.

At the time of decoding this system stream, the following scheme is devised on the system decoder side so that it can be reproduced seamlessly.

The system decoder is informed in advance of the STC discontinuity generation point stc1_PTS_end. Here, stc1_PTS which is a discontinuous point of STC
_End is a value obtained by adding the display time of the picture to the PTS of the last displayed picture of GOP-new-n on stc.

Then, the video decoder performs GOP-ne
After decoding wn, the following STC offset is calculated, and the DTS of the picture of GOP- (n + 1) is calculated.
When the STC offset value is equal to the STC value, the picture is decoded.

Here, the STC offset and the GOP-
The decoding time of the picture (n + 1) is given as follows.

STC offset = (GOP- (n + 1))
Pst of the picture displayed first) -stc1_P
TS_end GOP- (n + 1) picture decoding time = GOP
-DTS-STC offset of picture of (n + 1) When the value of stc1_PTS_end is equal to the value of STC, the picture is switched to the PTS of the picture displayed first in GOP- (n + 1). Decoding of a picture after this time is performed when DTS and STC of the picture become equal.

Next, the steps of the moving picture coding method will be described with reference to FIG.
Is shown in the flowchart of FIG.

In step S1, a GOP including an out point is input to the input MPEG-2 standard bit stream.
To the first GOP that can be decoded without predicting and referencing the pictures after the out point up to the out point according to the display order of the GOP. This is a reconstruction step of reconstructing a picture before the in point into a second GOP that can be decoded without predictive reference. Then, the process proceeds to the next step S2.

In step S2, a time delay required for at least the reconstruction in step S1 is performed on the input bit stream. In the next step S3, one of the bit stream from step S1 and the bit stream from step S2 is
The data is switched and output in units of OP. Then, this series of steps ends.

Next, the moving picture coding apparatus will be described. As shown in FIG. 6, the moving picture coding apparatus includes a pickup 12 for reading data recorded on an optical disc 101 as a recording medium, a demodulation circuit 13 for demodulating a signal from the pickup 12 into a bit stream, and a demodulation circuit 1.
And a buffer 1 for temporarily storing the bit stream from the H.3.

The pickup 12 reads out moving image data recorded on the optical disc 101 by irradiating a recording track on the signal recording surface of the optical disc 101 which is rotationally driven with laser light. The demodulation circuit 13 demodulates the moving image data provided from the pickup 12 into a bit stream of the MPEG2 standard. This input bit stream is temporarily stored in the buffer 1.

The moving picture coding apparatus further comprises a switch SW1 for switching the bit stream S21 from the buffer 1 to either the selected terminal A or the selected terminal B,
It has a decoder unit 2 for decoding the bit stream S21 from the selected terminal A of the switch SW1 into an image signal, and an encoder 4 for encoding the image signal S27 from the decoder unit 2.

The switch SW1 switches the bit stream S21 input from the buffer 1 to either the selected terminal A or the selected terminal B according to a control signal from the control circuit 11. The decoder unit 2 includes a switch SW
1 is decoded into an image signal S27, and the decoded image signal S27 is
Into a bit stream.

Here, the decoder section 2 includes a switch SW1
A switch 9 for switching the bit stream S22 from the selected terminal A to either the selected terminal A or the selected terminal B, and a buffer for temporarily storing the bit stream S23 from the selected terminal A of the switch 9 5 and a decoder 6 for decoding the bit stream from the buffer 5
And

The decoder section 2 includes a buffer 7 for temporarily storing the bit stream S24 from the selected terminal B of the switch 9, a decoder 8 for decoding the bit stream from the buffer 7, and a decoder 8 for decoding the bit stream from the buffer 7. A switch 10 for connecting the image output S25 to the selected terminal A and the decoded image output S26 from the decoder 8 to the selected terminal B, and switching to either the selected terminal A or the selected terminal B; I have.

The buffer 5 and the buffer 7 are memories for temporarily storing the bit stream input to the decoder unit 2. These buffers 5 and 7 are used when the decoders 6 and 8 at the subsequent stage decode the bit stream of the MPEG2 standard. The capacity of these buffers 5 and 7 is the MP2 of MPEG2.
＠Specified in the standard of ML (main profile / main level), for example, its capacity is 1.75
Mbit.

The bit streams supplied to the buffers 5 and 7 are switched by the switch 9 and supplied to one of the buffers 5 and 7.

The decoder 6 and the decoder 8 decode the bit stream compressed by the MPEG system and generate digital video data. The decoder 6 and the decoder 8 decode each picture according to decoding time management information (decoding timestamp; DTS) attached to each picture of the encoded data.

The decoder 2 is under the control of the control circuit 11. The control circuit 11 includes a buffer 5 and a buffer 7
Manages the bit occupancy of the bit stream and controls the bit rate of the bit stream to be read. Also, the control circuit 1
When performing skip reproduction, the switch 1 switches the switch SW1 at a timing of a break between a stream including an out-point picture and a stream including an in-point picture.

The decoder 2 has a bit stream S23.
And two decoders, that is, a decoder 6 and a decoder 8, so that the bit stream S24 and the bit stream S24 can be encoded simultaneously. The decoder 6 and the decoder 8 decode the bit stream for each GOP so as to connect the IN point and the OUT point in the bit stream.

The switch 10 switches the signals from the decoder 6 and the decoder 8 according to the timing of the control signal from the control circuit 11, and outputs the signals to the encoder 4.

Further, the moving picture coding apparatus comprises a switch S
A delay circuit 3 for delaying the bit stream S29 from the selected terminal B of W1 and an image signal S27 output from the switch 10 of the decoder unit 2 are converted to a bit stream S28.
And the bit stream S28 from the encoder 4 is input to a selected terminal A, and the bit stream S30 from the delay circuit 3 is input to a selected terminal B. One of these selected terminals A and B And a switch SW2 which is a switching means for switching to one side.

The delay circuit 3 sends the bit stream S29 from the selected terminal B of the switch SW1 to the selected terminal B of the switch SW2 with a predetermined delay. The delay time in this embodiment is 4 frame times as described later.

The decoder 2 and the encoder 4 constitute a re-encoding means for decoding an input bit stream and then encoding it. The encoder 4 re-encodes only the pictures necessary for the in-point and out-point out of the pictures decoded by the decoder unit 2.

Then, the moving picture coding apparatus operates the switch S
An interface circuit 14 for performing a predetermined conversion on the bit stream S41 from the W2, a switch SW1, a decoder unit 2, a delay circuit 3, an encoder 4, and a switch SW1
And a control circuit 11 for controlling

The interface circuit 214 performs an interface for converting the bit stream from the switch SW2 into data in the digital bus format of the IEEE 1394 standard and outputting the data to the outside.

Next, the signals of each part of the moving picture coding apparatus will be described with reference to FIG. In the buffer 5 of the decoder unit 2, as a bit stream S23, a sequence of pictures I02, B00, B01, P05, B0
3, B04, P08, B06, and B07 are input. This bit stream S23 is decoded by the decoder 6, and
This becomes the image signal S25. This image signal S25 includes frames F00, F01, F02, F03, and F04.

The picture coding type (picture-coding-code) is assigned to these frames as B, B, I, B, and P, corresponding to the type of the original picture. Here, the delay from the bit stream S23 to the image signal S25 is one frame, and the startup delay is zero.

Image signal S25 input from decoder 6
And the bit stream S28 (picture S25 input) obtained by encoding the picture coding type by the encoder 4 is represented by I
02, B00, B01, P04, and P03. This bit stream S28 is delayed by four frame times.

On the other hand, in the buffer 7, the pictures In2,
Bn0, Bn1, Pn5, Bn3, Bn4, Pn8, B
GOP that is an in-point side GOP composed of n6 and Bn
-N is input. This bit stream S24 is decoded by the decoder 8 and becomes an image signal. This image signal is output from frames Fn0, Fn1, Fn2, Fn3, and Fn.
4, Fn5, Fn6, Fn7, and Fn8.

[0134] Of these, the picture coding types are assigned to frames Fn5 to Fn8 as I, B, B, and P, corresponding to the type of the original picture. Here, the image signal S2 is converted from the bit stream S24.
The delay to 6 is 1 frame time, and the startup delay is 0.

Image signal S26 input from decoder 8
And a bit stream S28 (picture S26 input) obtained by encoding the picture coding type by the encoder 4 is represented by I
n5, Pn8, Bn6, and Bn8.
This bit stream S28 is delayed by four frame times.

GOP- (n + 1) is input to the delay circuit 3, delayed by 4 frames, and
It is output as 30.

Bit stream S2 from encoder 4
8 and the bit stream S30 from the delay circuit 3 are switched by the switch SW2 under the control of the control circuit 11, and GOP-new-0 after re-encoding GOP-0.
And GOP-n are re-encoded words GOP-new-n
I02, B00, B01, P04, B0
3, In5, Pn8, Bn6, Bn8 followed by the next GO
P- (n + 1) becomes the output bit stream S31.

Note that the present invention is not limited to the above embodiment. For example, a recording medium that supplies moving image data to the moving image encoding device is not limited to an optical disk, and the format of an output signal is not limited to the IEEE 1394 standard.

Next, a second embodiment of the present invention will be described. In the second embodiment, the present invention is applied to, for example, a moving image recording / reproducing apparatus 20 having the configuration shown in FIG.

When editing a bit stream already recorded on the recording medium 40 in units of pictures, the moving picture recording / reproducing apparatus 20 can seamlessly skip and reproduce the edit point (out-point picture to in-point picture). The video data is re-encoded and recorded on the recording medium 40 as described above.

Here, the out point picture (out point) refers to the last picture of one story when two stories are edited into one story.
The in-point picture (in-point) refers to the first picture of the other story of the two stories. That is, the video recording / reproducing apparatus 20 re-encodes the video data so that the video data can be reproduced seamlessly from the out point to the in point.

Specifically, the moving image recording / reproducing apparatus 20
Is a reading unit 21 that reads data from the recording medium 40.
An edit information input unit 22 for inputting information of a picture to be edited, a demodulation circuit 23 for performing demodulation processing, an error correction circuit 24 for performing error correction processing, and a buffer memory 25 for temporarily storing a bit stream. A demultiplexer 26 for separating a bit stream into a video stream and an audio stream; and an analysis unit 27.

When the user edits a video program recorded on the recording medium 40 and reproduces a new program, the edit information input unit 22 reads out the edit information of the out point and the in point, and reads the edit information of the out point and the in point. To the unit 27.

The reading section 21 supplies the data read from the recording medium 40 to the buffer memory 25 via the demodulation circuit 23 and the error correction circuit 24. Demultiplexer 2
6 reads out the bit stream stored in the buffer memory 25 and first supplies the bit stream stored in the buffer memory 25 to the analysis unit 27, and then separates the stream into a video stream and an audio stream. b, the audio stream is supplied to a multiplexer 32 described later.

The analysis unit 27 analyzes the bit stream from the demultiplexer 26, vbv_delay (a parameter indicating the initial state of the buffer at the time of random access; defined by the MPEG video picture header), and the like. Re-encoding method (picture
It changes the coding type and allocates the amount of coded bits at the time of re-encoding) and the re-multiplexing method, and controls the video encoder 30 and the multiplexer 32 according to the determined contents.

The video recording / reproducing apparatus 20 includes a video decoder 29 for decoding a video stream,
Video encoder 30 for encoding video data
A switching circuit 31 for switching and outputting a supply source video stream; a multiplexer 32 for multiplexing streams; and an error correction circuit 3 for adding an error correction code.
3 and a modulation circuit 34 for performing a modulation process on the bit stream.
And a writing unit 35 that writes the bit stream to the recording medium 40.

The video decoder 29 decodes the video stream and supplies video data to the video encoder 30. The video encoder 30 encodes video data at a bit rate according to the control of the analysis unit 27,
The video stream is supplied to a terminal a to the switching circuit 31.
The video decoder 29 and the video encoder 30
Is used when reconstructing a GOP including an IN point and an OUT point and GOPs before and after the GOP.

The switching circuit 31 is configured to be switchable to a terminal a or a terminal b, and is switched to the terminal a when re-encoding, and to the terminal b when not re-encoding. And
The switching circuit 31 outputs one of the video streams supplied to the terminal a and the terminal b, and
Feed to 2.

The multiplexer 32 performs a multiplexing process on the video stream from the switching circuit 31 and the audio stream from the demultiplexer 26 to generate a bit stream, and generates an error correction circuit 33 and a modulation circuit 3.
4 to the writing unit 35.

The writing section 35 records the bit stream from the modulation circuit 34 on the recording medium 40.

Here, when the out-point side video and the in-point side video are re-encoded and the re-encoded out-point side video is continuously played back to the re-encoded in-point side video, it is defined by MPEG. Care must be taken that the VBV buffer, which is the buffer of the virtual decoder, does not fail. Here, the fact that the VBV buffer does not fail means that the buffer overflows or underflows.
It does not flow.

Therefore, in the moving picture recording / reproducing apparatus 20 having such a configuration, the analysis unit 27 controls the video encoder 30 and the multiplexer 32 according to the flowchart shown in FIG. 9 to perform re-encoding and multiplexing processing.

The analysis unit 27 determines the VBV when decoding the first picture of the video on the in-point side to be re-encoded.
The initial bit occupancy Bvsd of the buffer is determined (step S11). The value of the initial bit occupation amount Bvsd may be set to a value suitable for an application. For example,
The maximum value of the VBV buffer (1.75 Mbit in the case of MPEG2MP @ ML) may be used. Then, the analysis unit 27 performs re-encoding and re-multiplexing of the video on the out-point side (step S12), and then performs re-encoding and multiplexing processing on the video on the in-point side. The multiplexer 32 is controlled (step S13).

The analyzing section 27 controls the re-encoding and multiplexing of the out point in step S12, specifically according to the flowchart shown in FIG.

In step S21, the analysis unit 27
The picture type of each picture of the video to be re-encoded is determined, and the process proceeds to step S22.

Here, for example, a video stream before re-encoding is configured as shown in FIG.
The recording medium 4 is displayed in the order of GOP-0, GOP-1,.
It is assumed that it is recorded as 0. The symbols I, P, and B represent I, P, and B pictures, respectively, and the subscript numbers of I, P, and B indicate the display order of the pictures in the GOP. For example, in GOP-0, the bit stream is configured in the order of I12, B10, B11,..., But the image actually displayed after decoding is in the order of B10, B11, I12,.
Re-encoding when the user designates B11 of GOP-1 as the out-point picture Pout will be described.

As for the determination of the picture type, for example,
When the process of step S21 is performed for the first time, GO
Image B is re-encoded only for P-1.
The picture type is determined as shown in FIG. 11B, with 10 as an I picture, image B11 as a P picture, and others as shown in FIG. After step S26 described later, step S21
Is performed, the picture type is determined for each image of each GOP such that GOP-0 and GOP-1 are re-encoded.

In step S22, the bit occupancy Bos of the VBV buffer when decoding the first picture of the video to be re-encoded is determined, and the flow advances to step S23. This value is, for example, the GO of the original stream.
It is calculated from the value of vbv_delay of the first coded picture I12 of P-1.

In step S23, the analysis unit 27
The allocated bit amount of each picture of the video to be re-encoded is determined, and the process proceeds to step S24.

In step S24, the analysis unit 27
Assuming that encoding is performed with the above-mentioned allocated bit amount, the bit occupation amount Bend1 of the virtual decoder after decoding is calculated in the last picture of the video to be re-encoded, and the process proceeds to step S25.

In step S25, the analysis unit 27
Bend1 determines whether or not the initial bit occupancy Bvsd or more,
When it is determined that it is not less than the initial bit occupancy Bvsd, the process proceeds to step S27, and when it is determined that it is not more than the initial bit occupancy Bvsd, the process proceeds to step S26.

In step S26, the analysis unit 27
The coded picture for starting the re-encoding is changed to the picture of the GOP before the out-point picture, and step S2
Return to 1. Then, in step S21, the analysis unit 27 determines the picture type as shown in FIG. 11C, and performs the processing in step S22 and subsequent steps in the same manner as described above.

In step S27, the analyzing unit 27
Controlling the video encoder 30 and the multiplexer 32 to re-encode and multiplex the video,
The process in step S12 described above ends. Note that the control contents of the analysis unit 27 are described in step S26.
Depends on whether or not the processing has been performed.

More specifically, the analyzing unit 27 determines in step S2
When the process proceeds to step S27 without passing through step S6, only GOP-1 including the out point is re-encoded, and FIG.
As shown in (B), the video encoder 30 is controlled to generate GOP-1-x. That is, the video encoder 30 converts the B10 of GOP-1 into the I picture of the I picture.
Re-encode to 10x, re-encode B11 of the out-point picture to P11x of the P-picture, and
Generate P-1-x. That is, the video encoder 30
Is the image after the image at the out point (for example, GOP
GOP including the out-point image B11 so that decoding can be performed without predictive reference to I12, B13, etc.
-1 is a GOP-1-x composed of images I10x and P11x
Is re-encoded. Note that the SQE (sequential
ence_end_code) indicates the end of the GOP.

At this time, the bit occupancy of the buffer of the virtual decoder changes as shown in FIG. Note that the vertical axis represents the bit occupancy of the buffer of the virtual decoder,
x is the maximum bit size of the buffer. The horizontal axis represents the passage of time. The time t (*) on the time axis indicates the decoding time of the coded picture. For example, t (P08) is P
08 indicates the decoding time. The slope of the straight line rising to the right represents the input bit rate to the buffer. For example, t (P0
From 8) to t (B06), the bit stream is input to the buffer, and the bit occupancy of the buffer increases. Then, at the decoding time of each encoded picture, the bit amount of the encoded picture is immediately removed from the buffer, and the bit occupancy of the buffer is reduced. The reduction of the bit occupancy at each decoding time differs depending on the picture type, and is different for I picture, P picture,
The decrease width is larger in the order of the B pictures.

In this case, as shown in FIG.
When the bit amount to be allocated to each of the pictures I10x and I11x of P-1-x is determined, the bit occupation amount Bend from the beginning.
1 is equal to or greater than the initial bit occupation amount Bvsd1, and it can be seen that GOP-1 may be re-encoded as it is.

When the flow proceeds to step S27 after step S26, the analyzing unit 27 can re-encode the out point in two modes as described below. Which mode to select depends on whether the maximum length of the GOP is limited by the application.

As a first mode, the video encoder 30
Re-encodes GOP-0 and GOP-1 into one GOP (GOP-Ax) as shown in FIG. 11 (C). In GOP-Ax, B from I02 of GOP-0
07 is directly copied to GOP-Ax. G
OP-1 B10 and B11 have been re-encoded into P-pictures P10x and P11x. That is, the video encoder 30 re-encodes each image of GOP-1 up to the out point into an inter-picture prediction coded image predicted from the GOP-0 image and a part of the GOP-1 image, The image (I13, P1) after the image at the out point
5) can be decoded without predictive reference to GOP-A-
x can be generated.

Here, assuming that GOP-0 is not re-encoded, the bit occupancy of the buffer of the virtual decoder changes as shown in FIG. 13, and Bend1 becomes smaller than the initial bit occupancy Bvsd. . That is, in step S23, each picture I10x, P of GOP-1-x is
When the allocated bit amount of 11x is determined, Bend1 <B
vsd. Therefore, the video encoder 30 changes the coded picture to start re-encoding to a picture of the GOP immediately before the GOP including the out-point picture. By re-encoding the original B11 into a P picture instead of an I picture, as shown in FIG.
The decrease in the bit occupancy of the buffer can be reduced, and as a result, Bend1 can be made equal to or greater than the initial bit occupancy Bvsd.

As a second mode, the video encoder 30
11 shows two GOPs of GOP-0 and GOP-1 in FIG.
As shown in (D), GOP-A-x1 and GOP-A-x
Re-encode to 2. In GOP-0, B from I02
04 is directly copied to GOP-A-x1. P08, B06 and B07 are IOPs of GOP-A-x2.
08x, B06x, and B07x. Further, in GOP-1, B10 and B11 are GO
Re-encoded to P10x and P11x of PA-x2. That is, the video encoder 30 re-encodes the GOP image including the out-point image into an inter-picture prediction coded image predicted from the GOP-0 image, and predicts and refers to an image subsequent to the out-point image. New GOP-A-x1 and GOP-A-x2 so that they can be decoded without
Can be generated. GOP-A-x1 and GOP-A-x2 are obtained by converting the above-described GOP-A-x into two GOPs.

The second mode is used when the maximum length of a GOP is limited. Here, GOP
Is limited to five pictures. That is, when the maximum length of the GOP is not limited, the processing may be performed in the first mode, and when the maximum length of the GOP is limited, the processing may be performed in the second mode.

As described above, the analyzing unit 27 calculates the bit occupancy
By controlling Bend1 not to fall below the initial bit occupancy Bvsd, to prevent the buffer from breaking down,
Overflow and underflow can be avoided.

Next, a case where the in-point is re-encoded will be described. The analysis unit 27 controls the re-encoding of the in-point according to the flowchart shown in FIG. 15 as the specific process of step S13 shown in FIG.

In step S31, the analysis unit 27
The picture type of the video to be encoded is determined, and the process proceeds to step S32. Here, for example, as shown in FIG. 16A, a video stream before re-encoding is configured and recorded on the recording medium 40 in the display order of GOP-n, GOP-m,. I do. As for the subscripts of I, P, and B, the subscript on the right side is G to which the image belongs.
OP, and the suffix on the left side indicates the order of images displayed after decoding in the GOP. For example, Bn3
Is a B picture belonging to GOP-n, and GOP-n
Is displayed third. Hereinafter, the user sets the P of GOP-n.
Re-encoding when n8 is designated as the in-point picture Pin will be described.

The picture type is determined by, for example,
When the process of step S31 is performed for the first time, GO
The image P is re-encoded only for P-n.
n8 as an I picture, image Bn6 as a B picture, image Bn7 as a B picture, and others as shown in FIG.
The picture type is determined as shown in FIG. Step S3
1 is not the first time, GOP-n and GOP-n
In order to re-encode m, a picture type is determined for each image of each GOP as shown in FIG.

In step S32, the analysis unit 27
The first picture of the video that is not re-encoded (eg, I
The initial bit occupation amount Bj of the VBV buffer when decoding m2) is calculated, and the flow advances to step S33. Here, since Pn8 is the in point, a GOP not re-encoded
Is GOP-m. Therefore, for Im2, which is the first picture of GOP-m, the initial bit occupancy Bj at the time of decoding is calculated. The initial bit occupancy Bj is calculated, for example, from the value of vbv_delay of Im2 of GOP-m.

In step S33, the analysis unit 27
The allocated bit amount of each picture of the video to be re-encoded is determined, and the process proceeds to step S34.

In step S34, the analysis unit 27
Assuming that encoding is performed with the determined allocated bit amount, the bit occupancy Bend2 of the virtual decoder after decoding of the last picture of the video to be re-encoded is calculated, and the process proceeds to step S35.

In step S35, the analysis unit 27
It is determined whether or not the bit occupancy Bend2 in step S34 is equal to or greater than the initial bit occupancy Bj.
Is greater than or equal to the initial bit occupation amount Bj, step S37
If the bit occupancy Bend2 is not equal to or greater than the initial bit occupancy Bj, the process proceeds to step S36.

In step S36, the analysis unit 27
The coded picture for which re-encoding is completed is changed to the picture of the next GOP after the in-point picture, and step S31 is performed.
Return to Then, the analyzing unit 27 determines in FIG.
The picture type shown in FIG. 6 (C) is determined, and the processing after step S32 is performed in the same manner as described above.

In step S37, the analysis section 27
Controlling the video encoder 30 and the multiplexer 32 to re-encode and multiplex the video,
The processing in step S13 described above ends. At this time, the control content of the analysis unit 27 differs depending on whether or not the processing in step S36 has been performed.

Specifically, when the process proceeds to step S37 without passing through step S36, the analysis unit 27 re-encodes only GOP-n including the in point, and
As shown in (B), the video encoder 30 is controlled to generate GOP-nx. At this time, the video encoder 30 sets Pn8, which is the in-point of GOP-n, to In8x of the I picture and Bn6 to Bn6x of the B picture.
Then, Bn7 is re-encoded to become Bn7x of the B picture to generate a new GOP-nx. That is, the video encoder 30 decodes the in-point image Pn so that the image before the in-point image (for example, Bn3, Bn4 or the like in GOP-n) can be decoded without predictive reference.
GOP-n including the image Nos. 8 as images In8x, Bn6x, Bn
Re-encode to GOP-nx consisting of 7x. Note that G
The SH (sequence_header) of OP-nx is the GO
It indicates the beginning of P.

At this time, the bit occupancy of the buffer of the virtual decoder changes as shown in FIG. in this case,
The bit occupancy Bend2 from the beginning is equal to or greater than the initial bit occupancy Bj, and even if only GOP-n is re-encoded, VBV
The buffer does not fail.

On the other hand, when the process proceeds to step S37 after step S36, the analyzing unit 27 controls the video encoder 30 to re-encode the out point in two modes as described below. Which mode to select depends on whether the maximum length of the GOP is limited by the application.

As the first mode, the video encoder 30
Re-encodes GOP-n and GOP-m into one GOP (GOP-Bx) as shown in FIG. More specifically, the video encoder 30 performs GOP-
n of Pn8, Bn6, Bn7 are In8x, Bn6x, B
n7x and re-encoded to GOP-m Im2
Is re-encoded to Pm2x, and the picture behind Im2 of GOP-m is re-encoded to the same picture type as the original.

Here, assuming that GOP-m is not re-encoded, the bit occupancy of the buffer of the virtual decoder changes as shown in FIG. 18, and Bend2 becomes smaller than the initial bit occupancy Bj. . That is, in the above step S32, each picture In8 of GOP-nx is
x, Bn6x, and Bn7x, when Bend2 <Bj, GOP-nx and GOP-
Re-encoding cannot be performed so that the trajectory of the bit occupancy of m is continuous. Thus, the video encoder 30 converts the original Im2 into a Pm that is a P picture instead of an I picture.
By re-encoding to 2x, as shown in FIG. 19, the required bit occupancy of the buffer can be reduced, and the re-encoding is performed so that the trajectory of the bit occupancy of the buffer on the in-point side is continuous. be able to.

As a second mode, the video encoder 30
FIG. 16 shows two GOPs of GOP-n and GOP-m.
As shown in (D), GOP-B-x1 and GOP-B-x
Re-encode to 2. The video encoder 30 is a GO
Pn of Pn8, Bn6, Bn7 is converted to In8x, Bn6
x, Bn7x, and GOP-m Im2,
Bm0 and Bm1 are re-encoded into Pm2x, Bm0x and Bm1x. Further, the video encoder 30
Pm5 of Pm is re-encoded into Im5x of GOP-B-x2, and pictures behind Pm5 of GOP-m are re-encoded to the same picture type as the original.

Note that GOP-B-x1 and GOP-B-
x2 corresponds to the above-mentioned GOP-Bx for two GOPs.

Whether the first mode or the second mode is selected depends on whether or not the out point is re-encoded, as in
When the maximum length of the GOP is not limited, the processing may be performed in the first mode. When the maximum length of the GOP is limited, the processing may be performed in the second mode.

As described above, the analysis unit 27 calculates the bit occupancy
By controlling Bend2 not to fall below the initial bit occupancy Bj, it is possible to prevent the buffer from breaking down.

As described above, the moving picture recording / reproducing apparatus 20
Means that if the video on the out-point side and the video on the in-point side are re-encoded, the buffer of the virtual decoder will fail when performing continuous playback from the re-encoded video on the out-point side to the re-encoded video on the in-point side Nothing can be played normally.

In the above-described embodiment, a case has been described where a program edited at the time of skip editing for designating an out point and an in point is created. However, the present invention can be applied to the case of insert editing. That is, in insert editing, first, an insert start point and an end point in a program to be inserted are designated, and a video sequence to be inserted next is newly encoded.
At this time, the re-encoding method of the out-point side video of this embodiment can be applied to the re-encoding of the video at the insertion start point portion of the program to be inserted. In addition, the re-encoding method of the out-point side video according to the present embodiment can be applied to the re-encoding of the video at the insertion end point portion of the inserted program.

Further, in the re-encoding processing of the in-point and the out-point described in the first and second embodiments, the re-encoding processing from the recording medium on which the moving image encoding control program for performing the above-described processing is recorded is performed. This may be performed by installing the program on a computer.

That is, the computer in which the above-mentioned control program is installed has a functional configuration as shown in FIG. 6 or FIG. 8, and when it is instructed to skip a part of the moving image program and reproduce it. Thus, the part before the skip start point and the part after the skip arrival point are efficiently encoded, and the program before the skip start point and the program after the skip arrival point can be seamlessly reproduced.

[0195]

As described above, the moving picture coding method, the moving picture coding apparatus and the recording medium according to the present invention provide an image of an image group including an out-point image for an input bit stream. Re-encode each image from the first display image of the image group to the out point into an inter-picture prediction coded image predicted from the image of the image group displayed immediately before the image group including the image at the out point. Then, a display image subsequent to the out-point image is reconstructed into an image group that can be decoded without predictive reference. Therefore, the part before the skip start point is efficiently encoded, and the program before the skip start point and the program after the skip arrival point can be seamlessly reproduced.

According to the moving picture coding method, the moving picture coding apparatus and the recording medium of the present invention, the first bit of the group of images displayed after the group of images including the image at the IN point is provided for the input bit stream. Is re-encoded into an inter-picture prediction coded image predicted from the images in the image group including the in-point image, and the display image prior to the in-point image is decoded without reference to prediction. And reconstruct them into a group of images that can be Therefore, the portion behind the skip end point can be efficiently encoded, and the program before the skip start point and the program behind the skip arrival point can be seamlessly reproduced. In addition, by grouping two image groups into one image group, for example, an I picture of an image group behind the two grouped images can be changed to a P picture, thereby reducing the amount of coding. be able to.

According to the moving picture coding method, the moving picture coding apparatus and the recording medium of the present invention, the input bit stream is processed in accordance with the display order in the group of images of the unit section including the first point. The image group up to point 1 and the image group of the previous unit section in accordance with the display order are reconstructed into a first image group that can be decoded without predicting and referring to the image behind the first point. At the same time, an image group from the second point according to the display order in the unit group image including the second point and an image group of the immediately following unit section according to the display order are placed in the second point. An earlier image is reconstructed into a second image group that can be decoded without predictive reference. Therefore, when it is specified that a part of a moving picture program is to be skipped and reproduced, a program before the skip start point and a program after the skip arrival point can be seamlessly reproduced, and the program before the skip start point can be reproduced. And the part behind the skip arrival point can be efficiently coded. In addition, by grouping two image groups into one image group, for example, an I picture of an image group behind the two grouped images can be changed to a P picture, thereby reducing the amount of coding. be able to.

The moving picture coding method, the moving picture coding apparatus and the recording medium according to the present invention, when the first point is a bidirectional predictive coded picture, form a group of pictures in a unit section including the first point. Decoding and re-encoding up to the first point according to the display order,
If the point is a forward prediction coded image or a bidirectional prediction coded image, the image group of the unit section including the second point is decoded, and the second and subsequent points are re-encoded according to the display order. . Therefore, when it is specified that a part of a moving picture program is to be skipped and reproduced, a program before the skip start point and a program after the skip arrival point can be seamlessly reproduced, and the program before the skip start point can be reproduced. And the part behind the skip arrival point can be efficiently coded.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating re-encoding of a bit stream.

FIG. 2 is a time chart showing a bit occupation amount of a VBV buffer.

FIG. 3 is a diagram illustrating a method for relaxing constraints on encoding conditions.

FIG. 4 is a diagram showing a system (multiplexed) stream.

FIG. 5 is a flowchart showing a series of steps of a moving picture encoding method.

FIG. 6 is a diagram schematically illustrating an overall configuration of a moving image encoding device.

FIG. 7 is a diagram illustrating an example of re-encoding of a GOP before and after an edit point.

FIG. 8 is a block diagram showing a configuration of a moving image recording / reproducing apparatus to which the present invention is applied in the second embodiment.

FIG. 9 is a flowchart illustrating a procedure for performing re-encoding and re-multiplexing of an out point and an in point in the moving image recording and reproducing apparatus.

FIG. 10 is a flowchart illustrating a procedure for re-encoding a video on the out-point side.

FIGS. 11A and 11B are diagrams for explaining an arrangement of encoded bit streams recorded on a recording medium; FIGS.
(D) is a diagram for explaining an arrangement of an encoded bit stream when an out point is edited by re-encoding.

FIG. 12 is a diagram illustrating the bit occupancy of a virtual buffer when edited by re-encoding.

FIG. 13 is a diagram illustrating the bit occupancy of the buffer of the virtual decoder when it is assumed that B11 of GOP-0 is re-encoded as I10x.

FIG. 14 is a diagram illustrating the bit occupancy of a virtual buffer when edited by re-encoding.

FIG. 15 is a flowchart illustrating a procedure for re-encoding an in point in the moving image recording / reproducing apparatus.

FIG. 16A is a diagram for explaining an arrangement of coded bit streams recorded on a recording medium, and FIGS.
(D) is a diagram for explaining an arrangement of an encoded bit stream when the in point is re-encoded.

FIG. 17 is a diagram illustrating the bit occupancy of the virtual buffer when the IN point is re-encoded.

FIG. 18 is a diagram illustrating the bit occupancy of the buffer of the virtual decoder when it is assumed that Im2 of GOP-m is re-encoded as Im2x.

FIG. 19 is a diagram illustrating the bit occupancy of the virtual buffer when the IN point is re-encoded.

[Explanation of symbols]

2 Decoder section, 4 encoder, 3 delay section, Pou
t out point picture, Pin in point picture, SW
1, SW2 switch, 20 moving image recording / reproducing device, 2
2 edit information input section, 26 demultiplexer, 29
Video decoder, 30 video encoder, 31 switching circuit, 32 multiplexer

Claims (30)

[Claims] 1. A moving picture coded bit stream having a plurality of pictures as a unit section, starting from an intra coded picture and subsequently being followed by an inter prediction coded picture, is output according to a display order. In the moving picture coding method for outputting a moving picture signal up to a point image, the input bit stream is an image of an image group including the image of the out point, and the image is output from the first display image of the image group. Each image up to the point is re-encoded into an inter-image predicted coded image predicted from the image of the image group displayed immediately before the image group including the out-point image,
A moving picture coding method comprising a reconstructing step of reconstructing a display picture subsequent to the out-point picture into a group of pictures that can be decoded without predictive reference. 2. The image processing apparatus according to claim 1, wherein the reconstructing step predicts a first intra-coded image of an image group including the out-point image from an image of a preceding image group of the image group including the out-point image. 2. The moving picture coding method according to claim 1, wherein the moving picture is re-encoded into a forward prediction coded image. 3. The method according to claim 1, wherein the reconstructing step comprises: determining an initial bit occupancy required for a virtual buffer of the decoder when decoding the first coded picture of the coded video signal displayed next to the out-point image. Calculated, is an image of an image group including the image at the out point, and each image from the first display image of the image group to the out point is re-encoded into one image group by an allocated bit amount. Assuming that encoding is performed, the last image to be encoded is decoded, and the bit occupancy at the end of decoding of the virtual decoder after decoding is calculated. The bit occupancy at the end of decoding is equal to or greater than the initial bit occupancy. When the images of the image group including the image at the out point are re-encoded into one image group, each image from the first display image of the image group to the out point is recoded. When the bit occupancy at the end of decoding is not equal to or greater than the initial bit occupancy, the inter-picture prediction coding predicted from the image of the image group displayed immediately before the image group including the out-point image 2. The image is re-encoded into an image.
The moving picture coding method according to the above. 4. A moving picture coded bit stream that starts with an intra-coded picture and is followed by an inter-picture predictive coded picture and has a unit section consisting of a plurality of pictures is output according to a display order. In a moving picture coding apparatus for outputting a moving picture signal up to a point image, an image of an image group including the image at the out point is output from the first display image of the image group with respect to the input bit stream. Each image up to the point is re-encoded into an inter-image predicted coded image predicted from the image of the image group displayed immediately before the image group including the out-point image,
A moving picture encoding apparatus, comprising: a reconstructing unit that reconstructs a display image subsequent to the out-point image into a group of images that can be decoded without predictive reference. 5. The image processing apparatus according to claim 1, wherein the reconstructing unit predicts a first intra-coded image of an image group including the out-point image from an image of a preceding image group of the image group including the out-point image. 5. The moving picture coding apparatus according to claim 4, wherein the moving picture coding apparatus re-encodes the forward predicted encoded image. 6. The reconstructing means determines an initial bit occupancy required for a virtual buffer of a decoder when decoding a first coded picture of a coded moving picture signal displayed next to the out-point image. Calculate and encode the images of the image group including the image at the out point, with the allocated bit amount when each image from the first display image of the image group to the out point is re-encoded into one image group Assuming that, when the last image to be encoded is decoded, the bit occupancy at the end of decoding of the virtual decoder after decoding is calculated, and the bit occupancy at the end of decoding is equal to or more than the initial bit occupancy. Is an image of an image group including the out-point image, and re-encodes each image from the first display image of the image group to the out-point into one image group. When the bit occupancy at the end of decoding is not equal to or greater than the initial bit occupancy, the inter-picture prediction coding predicted from the image of the image group displayed immediately before the image group including the out-point image 5. The moving picture coding apparatus according to claim 4, wherein the moving picture is re-encoded into an image. 7. A moving picture coded bit stream that starts with an intra-coded picture and is followed by an inter-picture predictive coded picture and has a unit group of a plurality of pictures as a unit section is output according to a display order. In a recording medium on which a control program for outputting a moving image signal up to a point image is recorded, for an input bit stream, an image of an image group including the image at the out point, and the first display image of the image group From the respective images up to the out point, re-encoding into an inter-picture prediction coded image predicted from the image of the image group displayed immediately before the image group including the out point image,
A recording medium on which a control program for reconstructing a display image subsequent to the out-point image into a group of images that can be decoded without predictive reference is recorded. 8. A forward prediction code predicted from an image of an image group immediately before the image group including the out-point image, wherein the first intra-coded image of the image group including the out-point image is predicted. 8. The recording medium according to claim 7, wherein a control program for re-encoding the encoded image is recorded. 9. An initial bit occupancy required for a virtual buffer of a decoder when decoding a first coded picture of a moving picture coded signal displayed subsequent to the out-point image is calculated, and When it is assumed that each image from the first display image of the image group to the above-mentioned out point is encoded with the assigned bit amount when re-encoding into one image group, Decoding the last image to be encoded, calculating the bit occupancy at the end of decoding of the virtual decoder after decoding is completed, and when the bit occupancy at the end of decoding is greater than or equal to the initial bit occupancy, the out-point is calculated. Each of the images from the first display image of the image group to the out point is re-encoded into one image group, and the decoding If the end bit occupancy is not equal to or greater than the initial bit occupancy, the inter-picture prediction coding image predicted from the image of the image group displayed immediately before the image group including the out-point image is re-encoded. The recording medium according to claim 7, wherein a control program to be converted is recorded. 10. A moving picture coded bit stream having a unit group consisting of a plurality of pictures starting from an intra coded picture and subsequently being followed by an inter prediction coded picture according to a display order. In a moving image encoding method for outputting an image signal from an image, for an input bit stream, a first intra-encoded image of an image group displayed after the image group including the in-point image is obtained. Re-encoding to an inter-picture prediction coded image predicted from an image of an image group including the in-point image, and to a group of images that can be decoded without predicting and referring to a display image preceding the in-point image. A moving picture coding method comprising a reconstructing step of reconstructing. 11. The image processing apparatus according to claim 1, wherein the reconstructing step comprises: converting the first intra-coded image of an image group displayed after the image group including the in-point image into an image group including the in-point image. The moving picture coding method according to claim 10, wherein the moving picture is recoded into a forward prediction coded picture predicted from the picture. 12. The moving image according to claim 10, wherein the reconstructing step re-encodes the first forward prediction image after the in-point image in display order into an intra-coded image. Image coding method. 13. The reconstructing step comprises the steps of: initial bit required for a buffer of a decoder when decoding the first coded picture of the next succeeding image group in accordance with the display order of the image group including the in-point image. Calculating the occupancy, and re-encoding each image from the in-point image to the last display image of the image group including the in-point image into one image group. When it is assumed that encoding is performed with the allocated bit amount, the last image to be encoded is decoded, and the bit occupancy at the end of decoding of the virtual decoder after decoding is calculated. If the amount is equal to or greater than the amount, the images of the image group including the image at the in-point, and each image from the in-point of the image group to the last display image is re-encoded into one image group. If the bit occupancy at the end of the decoding is not equal to or greater than the initial bit occupancy, the first intra-coded image of the image group displayed immediately after the image group including the image at the in-point is replaced with the input image. The moving image encoding method according to claim 10, wherein the image is re-encoded into an inter-image prediction encoded image predicted from an image of an image group including a point image. 14. A moving picture coded bit stream having a unit section of an image group consisting of a plurality of pictures which start with an intra-coded picture and subsequently follow an inter-picture coded picture, according to a display order. A video encoding apparatus for outputting an image signal from an image of a point, wherein, for an input bit stream, the first intra-encoding of an image group displayed one after the image group including the image of the in-point An image that can be re-encoded into an inter-picture prediction coded image predicted from an image of an image group including the in-point image, and a display image preceding the in-point image can be decoded without predictive reference. A moving picture coding apparatus comprising reconstructing means for reconstructing a group. 15. The image processing apparatus according to claim 1, wherein the reconstructing unit converts the first intra-coded image of an image group displayed after the image group including the in-point image into an image group of the image group including the in-point image. 15. The moving picture coding apparatus according to claim 14, wherein the moving picture coding apparatus re-encodes the picture into a forward prediction coded picture predicted from the picture. 16. The moving image according to claim 14, wherein said reconstructing means re-encodes the first forward prediction image after the in-point image in display order into an intra-coded image. Image coding device. 17. The decoding method according to claim 17, wherein said reconstructing means decodes an initial bit necessary for a buffer of a decoder when decoding a first coded picture of an immediately succeeding image group in accordance with a display order of the image group including the in-point image. Calculating the occupancy, and re-encoding each image from the in-point image to the last display image of the image group including the in-point image into one image group. When it is assumed that encoding is performed with the allocated bit amount, the last image to be encoded is decoded, and the bit occupancy at the end of decoding of the virtual decoder after decoding is calculated. If the amount is equal to or greater than the amount, the images of the image group including the image at the in-point, and each image from the in-point of the image group to the last display image is re-encoded into one image group. If the bit occupancy at the end of the decoding is not equal to or greater than the initial bit occupancy, the first intra-coded image of the image group displayed immediately after the image group including the image at the in-point is replaced with the input image. 15. The moving picture coding apparatus according to claim 14, wherein the moving picture coding apparatus re-codes an inter-picture prediction coding picture predicted from an image of a group of pictures including a point picture. 18. A moving picture coded bit stream having a unit section consisting of an image group consisting of a plurality of pictures which start with an intra-picture coded picture and subsequently follow an inter-picture coded picture, according to the display order. In a recording medium on which a control program for outputting an image signal from a point image is recorded, for an input bit stream, the first image of an image group displayed after the image group including the in-point image The inner coded image is re-encoded into an inter-image predicted coded image predicted from the image of the image group including the in-point image, and the display image prior to the in-point image is decoded without prediction and reference. A recording medium on which a control program for reconstructing a group of images that can be reproduced is recorded. 19. The first intra-coded image of an image group displayed one after the image group including the in-point image,
19. The recording medium according to claim 18, wherein a control program for re-encoding a forward prediction coded image predicted from an image of an image group including the in-point image is recorded. 20. A control program for re-encoding a first forward prediction image after an in-point image in a display order into an intra-coded image.
8. The recording medium according to 8. 21. calculating the initial bit occupancy required for the buffer of the decoder when decoding the first coded picture of the immediately succeeding image group according to the display order of the image group including the in-point image; The images of the image group including the image of the in-point, and each image from the image at the in-point to the last display image of the image group is encoded with the assigned bit amount when re-encoding into one image group. Assuming that the last image to be encoded is decoded, the decoding end bit occupancy of the virtual decoder after decoding is calculated, and when the decoding end bit occupancy is equal to or greater than the initial bit occupancy, Re-encoding each image from the in-point image of the image group including the in-point image to the last display image into one image group, and completing the decoding When the time bit occupancy is not equal to or more than the initial bit occupancy, the first intra-coded image of the image group displayed one behind the image group including the in-point image includes the in-point image. 19. The recording medium according to claim 18, wherein a control program for re-encoding an inter-picture prediction coded image predicted from the images of the image group is recorded. 22. A bit stream obtained by encoding an image signal in which an intra-coded image and a predicted coded image predicted based on the image are used as elements, and an image group including a plurality of images is used as a unit section. In contrast, in a moving picture coding method for outputting a signal skipped from a first point to a second point according to a display order, an image of a unit section including the first point is provided for an input bit stream. The first of the groups according to the display order
Are reconstructed into a first image group that can be decoded without predicting and referring to an image subsequent to the first point, with the image group up to the point and the previous unit section according to the display order. At the same time, in the image group of the unit section including the second point, the image group from the second point according to the display order and the image group of the next unit section according to the display order are referred to as the image group of the second section. A moving image coding method, comprising: a reconstructing step of reconstructing an image before point 2 into a second image group that can be decoded without predictive reference. 23. A predictive coded image predicting in a forward direction according to a display order of the images,
23. The moving picture coding method according to claim 22, comprising a bidirectional prediction coded picture predicted bidirectionally according to a display order. 24. The moving picture coding method according to claim 22, further comprising delay means for delaying an input bit stream by at least a time required for said reconstruction step. 25. A bit stream obtained by encoding an image signal in which an intra-coded image and a predicted coded image predicted based on the image are used as elements, and an image group including a plurality of images is used as a unit section. A video encoding device that outputs a signal skipped from a first point to a second point in accordance with the display order. The first of the groups according to the display order
Are reconstructed into a first image group that can be decoded without predicting and referring to an image subsequent to the first point, with the image group up to the point and the previous unit section according to the display order. At the same time, in the image group of the unit section including the second point, the image group from the second point according to the display order and the image group of the next unit section according to the display order are referred to as the image group of the second section. A moving picture coding apparatus comprising: reconstructing means for reconstructing a picture prior to point 2 into a second picture group that can be decoded without predictive reference. 26. A bit stream obtained by encoding an image signal in which an intra-coded image and a predicted coded image predicted based on the image are used as elements, and an image group including a plurality of images is used as a unit section. On the other hand, in a recording medium on which a control program for outputting a signal skipped from a first point to a second point according to a display order is recorded, a unit including the above-mentioned first point is used for an input bit stream. The first image according to the display order in the image group of the section.
Are reconstructed into a first image group that can be decoded without predicting and referring to an image subsequent to the first point, with the image group up to the point and the previous unit section according to the display order. At the same time, in the image group of the unit section including the second point, the image group from the second point according to the display order and the image group of the next unit section according to the display order are referred to as the image group of the second section. A recording medium on which a control program for reconstructing an image before point 2 into a second image group that can be decoded without predictive reference is recorded. 27. An intra-coded image, a forward predicted coded image predicted in the forward direction according to the display order of the image, and a bidirectional predicted coded image predicted bidirectionally in the display order. A moving image code for outputting a signal skipped from a first point to a second point according to a display order with respect to a bit stream obtained by encoding an image signal having an image group including a plurality of images as a unit interval. In the encoding method, when the first point is a bidirectional predictive coded image, a group of images including a unit of the first point is decoded, and up to the first point according to the display order is re-encoded. If the second point is a forward predictive coded image or a bidirectional predictive coded image, the image group of the unit section including the second point is decoded, and the second group is decoded according to the display order. Having a reconstruction step to re-encode points and beyond
A moving picture coding method characterized by the following. 28. The moving picture coding method according to claim 27, further comprising delay means for delaying an input bit stream by at least a time necessary for the reconstruction step. 29. An intra-coded image, a forward predicted coded image predicted in the forward direction according to the display order of the images, and a bidirectional predicted coded image predicted bi-directionally in the display order. A moving image code for outputting a signal skipped from a first point to a second point according to a display order with respect to a bit stream obtained by encoding an image signal having an image group including a plurality of images as a unit interval. In the coding apparatus, when the first point is a bidirectional predictive coded image, a group of images in a unit section including the first point is decoded, and up to the first point according to a display order is re-encoded. When the second point is a forward prediction coded image or a bidirectional prediction coded image, the image group of the unit section including the second point is decoded, and the second group is decoded according to the display order. Operation comprising reconstructing means for re-encoding the points after the point. Image encoding device. 30. An intra-coded image, a forward predicted coded image predicted in the forward direction according to the display order of the image, and a bidirectional predicted coded image predicted bi-directionally in the display order. A control program for outputting a signal skipped from a first point to a second point in accordance with a display order with respect to a bit stream obtained by encoding an image signal having an image group including a plurality of images as a unit interval. In the recorded recording medium, when the first point is a bidirectional predictive coded image, a group of images in a unit section including the first point is decoded and up to the first point according to the display order. If the second point is a forward predictive coded image or a bidirectional predictive coded image, decodes the image group of the unit section including the second point and follows the display order. A control program for re-encoding the second and subsequent points is described. Recording medium, characterized in that it is.