JPH08102913A - Method for reverse production of compressed moving picture data - Google Patents

Abstract

PURPOSE: To obtain a method for reproducing compressed moving picture data through the use of a frame memory of a smaller storage capacity when the data of the MPEG system are reversely reproduced. CONSTITUTION: The compressed moving picture data of the MPEG system in the reverse reproduction are received in the reverse order from the forward reproduction in the unit of the GOP. Switch circuits 8, 9, 5 apply switch control to an input series and I-and P-pictures of 1 GOP and an I-picture relating to a preceding GOP in the forward direction are decoded and reproduced and the result is stored in a frame memory 2a, and a B picture of 1 GOP is stored in a data memory 7 without converting the coded data of the B-picture. While the data in the memories 7, 2a are being rewritten by the FILO system, the B-picture in the data memory 7 are decoded and reproduced and a switch circuit 6 is switch-controlled to provide an output of reproduced image data of each picture continuously in the reverse order.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a reverse reproduction method of compressed moving image data, and relates to an MPEG (Moving Picture Image Co
ding Experts Group) method for playing back compressed moving image data in the backward direction using a frame memory having a small storage capacity.

[0002]

2. Description of the Related Art M is used as an encoding method for compressed moving image data.
The PEG system has been standardized, and in particular MPEG2, which became an international standard in November 1994, is HDTV (High Definition Televisio).
It is drawing attention as a method that can compress image data of higher quality than current television broadcasting such as n) without degrading the image quality. The MPEG video coding algorithm adopts a hybrid coding method that combines motion-compensated interframe prediction and discrete cosine transform (DCT), but GOP (Group o) to enable a random access function.
f Pictures) are encoded as a unit.

Here, each GPO is a frame that is motion-compensated and predicted from an I picture that is intra-frame coded image data and does not use motion compensation, and the immediately preceding reproduced image data (I picture or P picture). P that has been inter-coded
Each picture is composed of a picture and a B picture that has been subjected to inter-frame coding of bidirectional prediction by motion-compensated prediction from previous and subsequent reproduced image data (I picture or P picture). 6A is arranged in the order as shown in FIG. 6A, but after being encoded on the recording medium,
They are arranged in the order shown in (B). That is, B0 to P14 in the data sequence of the original image form one GPO, and the pictures of the subsequent GOPs in FIGS. 6A and 6B are also periodically arranged in the same order. still,
The numbers attached to the symbols (I, P, B) indicating each picture indicate the order in the case of forward reproduction and output, and the order is naturally in accordance with the data series of the original image data.

FIG. 7 shows a functional block circuit diagram of a reproducing apparatus for reproducing MPEG-coded moving image data. In the figure, 1 is a decoder for variable-length decoding / inverse quantization / inverse DCT of input image data, 2 is a frame memory, 3 is a motion compensation circuit, 4 is an adder, 5 is a frame memory 2 The switch circuit that is set to ON when writing data to
Is a switch circuit connected to the a side when outputting the data after addition by the adder 4 and connected to the b side when outputting the data of the frame memory 2, The compressed video data is decoded and reproduced in the reverse order.

When the data series on the recording medium shown in FIG. 6B is reproduced in the forward direction, the data series are input to the reproducing apparatus in the order of decoding and decoding of each picture. The reproduction is executed according to the procedure shown in FIG. 8, which will be described below step by step. However, in FIG. 8, ○ is a decoder
The decoding state at 1, □ indicates the storage state in the frame memory 2, the solid line with an arrow indicates the output direction, and the dotted line with an arrow indicates the inter-frame prediction relationship. It should be noted that the arrow side of the dotted line with an arrow means that decoding is performed by referring to the decoded image data on the start point side.

(1) The I2 picture is decoded by the decoder 1 to obtain reproduced image data, and the reproduced image data is written into the frame memory 2 via the switch circuit 5 set to the ON state. (2) B0 and B1 pictures are input, but they are I2
It cannot be played back unless the previous P picture is referenced,
In the first GOP, the P picture is not reproduced because it does not exist. (3) The P1 picture is decoded by the decoder 1, but the decoder 1 outputs the decoded difference image data to the adder 4, and also outputs the frame selection data to the frame memory 2 to reproduce the I2 picture. The image data is read out to the motion compensation circuit 3, and the motion vector obtained from the P5 picture is output to the motion compensation circuit 3. Then, the differential image data relating to the P5 picture and the predicted image data obtained by the motion compensation circuit 3 are added by the adder 4 to reproduce the image data of the P5 picture, and the reproduced image data is set to the ON state in the switch circuit. Write to frame memory 2 via 5. (4) For the B3 and B4 pictures to be input next, the reproduction procedure is the same as that for the P5 picture, but B3 and B4
Since the picture is bidirectional prediction inter-frame coded data, the I2 picture and P5 picture previously stored in the frame memory 2 are used as reference image data. That is, the motion compensation circuit 3 creates each predicted image data by using two motion vectors and two reproduced image data, and the adder
4 adds the difference image data relating to the B3 picture and the B4 picture and the created prediction image data to obtain the reproduced image data of the B3 picture and the B4 picture. However, the reproduced image data of the B3 picture and the B4 picture are not written in the frame memory 2, but the switch circuit 6 is connected to the a side to be output.

Thereafter, as will be understood from FIG. 6B and FIG. 8, when the pictures of P8, P11, P14, ... Are reproduced, the reproduced image data of P5, P8, P11 ,. , Respectively (B6 and B7), (B9 and B10), ..., (B15 and B16)
When reproducing each picture, the reproduced image data of (P5 and P8), (P8 and P11), ..., (P14 and I17) are sequentially referred to. The switch circuit 6 outputs B as the output of the adder 4.
When the reproduced image data of the picture is obtained, it is connected to the side a, and when the referenced I picture or P picture of the frame memory 2 is output, it is connected to the side b. The reproduced image data of each picture is output in the order of numbers 2 and subsequent numbers attached to the symbols (I, P, B) indicating each picture in FIG.

[0008]

By the way, when the recorded data sequence is reproduced in the backward direction, the output order of the reproduced image data may be reversed as a result, which is the case of the forward reproduction. It is necessary to perform decoding / reproduction in units as in the case of forward reproduction. This is because, as described above, each GPO cannot completely reproduce the P picture or B picture in the GOP unless it is reproduced using the I picture, which is the intra-frame coded image data, as the basic picture.

Therefore, in the conventional reproducing apparatus, in consideration of the case of backward reproduction, a frame memory 2 having a storage capacity capable of storing reproduced image data of 1 GOP is used, and each frame memory 2 is divided into GOP units. All image data obtained by decoding and reproducing a picture is stored, and the output order is reversed from that in the case of forward reproduction, and each image data is output. Therefore, in the case of forward reproduction, it is sufficient that the frame memory 2 has a capacity capable of storing reproduced image data for two frames, whereas in the case of reverse reproduction, 1 GOP It is necessary to have the capacity to store all the reproduced image data in
In a Video-CD or the like, 1 GOP is composed of 15 pictures as shown in FIG. 6, so the required storage capacity of the frame memory 2 is 15 frames. That is, it is necessary to prepare a large-capacity frame memory 2 for backward reproduction with a relatively low operation frequency.

Therefore, the present invention provides a reproducing method which enables backward reproduction by using a frame memory having a smaller storage capacity in a reproducing apparatus for compressed moving image data of the MPEG system, thereby reducing the cost of the reproducing apparatus. It was created for the purpose of making it easier.

[0011]

According to the present invention, there is provided a decoding means for performing variable length decoding / inverse quantization / inverse discrete cosine transform on I picture, B picture and P picture which are compressed moving picture data, and I picture as described above. The I-picture reproduction data decoded by the decoding means is stored, and the P-picture is decoded by the decoding means and then the motion compensation processing is performed by the motion compensation means, which will be described later.
First storage means for storing picture reproduction data, and P picture decoded data decoded by the decoding means for reproducing the I picture or P picture immediately preceding the P picture by time, and the first storage. The motion compensation processing is performed using the I picture reproduction data or the P picture reproduction data stored in the means, and B decoded by the decoding means.
I picture reproduction data or P picture reproduction stored in the first storage means by reproducing I picture or P picture, which is one before and one time after the picture decoded data, in time. In a reproducing apparatus for compressed moving image data including a motion compensating unit that performs a motion compensating process using data, a second storage unit is provided before the decoding unit, and the compressed moving image data is reproduced in the backward direction. In the first storage means, reproduction data of I and P pictures in one GOP and I picture reproduction data of the previous GOP in the forward direction are stored in the first storage means, and in the second storage means 1GO.
Each B picture in P is FILO (First-in Last
-out) method while sequentially rewriting and storing, and decoding the B picture read from the second storage means by the decoding means,
The reproduction data of the I picture or P picture to be referred to for the decoded B picture decoded data is the first
Reverse reproduction of compressed moving image data, characterized in that the B picture decoded data is read from the storage means, subjected to motion compensation processing on the B picture decoded data, reproduced, and reproduced data of each picture is output in reverse reproduction sequence. According to the method.

[0012]

In the case of backward reproduction, compressed moving image data is GOP
Each GO is input in the reverse order of the case of forward playback.
The pictures in P are input in forward order. In the present invention, 1 GOP worth of undecoded B pictures corresponding to 1 GOP are sequentially rewritten and stored in the second storage means, and 1 GOP worth of decoded / reproduced I pictures are stored in the first storage means.
The reproduction data of the picture and the P picture and the I picture reproduction data of the immediately preceding GOP in the forward direction are stored while being sequentially rewritten by the FILO method.

Therefore, the reproduction data of the I picture and P picture to be referred to when reproducing each B picture in the second storage means is stored in advance in the first storage means, and at the timing of decoding each B picture. Each B picture can be reproduced by reading the corresponding I picture or P picture from the first storage means to the motion compensation means.

In this case, it is not necessary for the first storage means to store the reproduction data of the B picture in the GOP, it is sufficient to store only the series of I pictures and P pictures, and the storage capacity can be set small. In addition, the second storage means is 1G before decoding.
Although B pictures for OP are stored, the amount of data is small and the storage capacity is small because the data is before decoding.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The embodiments of the "reverse reproduction method of compressed moving image data" of the present invention will be described in detail below with reference to FIGS. First, Fig. 1 shows the video data encoded by Video-CD.
The functional block circuit diagram of the reproducing | regenerating apparatus which reproduce | reproduces (MPEG data) is shown. In the figure, the circuit configuration after the decoder 1 is the same as that in the case of FIG. 6, but the frame memory 2a has a storage capacity equivalent to 6 frames, and the reproduction of FIG. The frame memory 2 in the device has to have a storage capacity corresponding to 15 frames, whereas it has a small storage capacity. Further, the reproducing apparatus of the present embodiment has a data memory 7 on the input side of the decoder 1.
And two switch circuits 8 and 9 are provided, the input data can be directly input to the decoder 1 by connecting the switch circuits 8 and 9 to the side a, and the switch circuits 8 and 9 to the side b. The feature is that input data can be read / written to the data memory 7 by connecting.

When performing forward reproduction using this reproducing apparatus, each picture is decoded and reproduced by the procedure described in the above-mentioned prior art, and reproduced in the reproduction order shown in FIG. 6 (A). Output image data. In that case, as shown in FIG. 8, the frame memory 2 is moved forward and backward [I picture and P
It is sufficient to store the reproduced image data of [picture] or [P picture and P picture], and only the area corresponding to the reproduced image data of 2 pictures is used.

On the other hand, when performing reverse reproduction, each picture is decoded and reproduced by using the data memory 7 and the frame memory 2a by the following method, and reproduced image data is reversely reproduced and output. However, in the present embodiment, a procedure for reproducing backward the compressed moving image data of the MPEG system arranged in the forward direction on the recording medium as shown in FIG. 2 will be described. Here, the compressed moving image data of FIG. 2 is the same as that shown in FIG. 6, but in the case of backward reproduction, the GOP units are input in the reverse order, and the pictures in each GOP are in the forward direction. Will be entered in the order of.

When a backward reproduction instruction is made while B37 on the recording medium is being reproduced in the forward direction, << I32 → B30 → B31 → P35 → B33 → B34 → P38 from the recording medium.
→ B36 >> → I17 → B15 → B16 → P20 → B18 → B19 → P
23 → B21 → B22 → P26 → B24 → B25 → P29 → B27 → B2
8 >> → << I2 → B0 → B1 → P5 → B3 → B4 → P8 → B6 → B7
Each picture is read in the order of → P11 → B9 → B10 → P14 → B12 → B13 >> →, and is input to the reproducing apparatus. still,""
In the figure, the GOP is shown, but the GOP that was read first was being played back at the time when the backward playback instruction was issued.
It is up to B36 just before 37.

With respect to the above-mentioned input series data, the reproducing apparatus executes the following procedure for each GOP. (1) When inputting an I picture, set the switch circuits 8 and 9 to the connection state on the side a and the switch circuit 5 to the ON state, and write the I picture reproduction data decoded by the decoder 1 to the frame memory 2a. Put in. (2) At the time of inputting a P picture, the connection state of the switch circuits 8, 9 and 5 is the same as in the case of (1), and the P picture is decoded by the decoder 1
The differential image data decoded in step 1 is output to the adder 4, the decoder 1 outputs the motion vector data related to the P picture to the motion compensation circuit 3, and the frame selection data is output to the frame memory 2a to reproduce the I picture. The data (in the case of the first P picture in the GOP) or the P picture reproduction data temporally before the P picture input this time is read to the motion compensation circuit 3, and the predicted image data obtained by the motion compensation circuit 3 The reproduced image data of the P picture obtained by adding the decoded differential image data by the adder 4 is written to the frame memory 2a via the switch circuit 5. (3) When a B picture is input, the switch circuits 8 and 9 are set to the connection state on the b side, and the B picture in 1 GOP is written to the data memory 7 as it is.

In each of the above procedures, the I picture in the GOP first input to the frame memory 2a, the reproduced image data of each P picture, and the I related to the previous GOP in the forward sequence.
The picture (I17) is stored, and each B picture in the first GOP is stored in the data memory 7 as encoded data. At that stage, the reproduction procedure of the B picture and the output of the reproduced image data are output. Move to procedure.

The specific reproducing procedure is shown in FIG. 3, which will be described below in order with reference to FIG. However,
The symbols ○, □, and the solid and dotted lines with arrows have the meaning shown in FIG.
However, the symbol Δ indicates the storage state in the data memory. First, the switch circuits 8 and 9 are set to the connection state on the side b, and the last input in the GOP is the data memory 7
The B pictures stored in B are sequentially read to the decoder 1 and decoded, and the decoder 1 outputs the motion vector data related to the B picture to the motion compensation circuit 3 and the frame selection data to the frame memory 2a. The reproduced image data to be referred to is read out to the motion compensation circuit 3, and the predicted image data of the B picture obtained by the motion compensation circuit 3 and the difference image data after decoding are added by the adder 4 to the reproduced image data of the B picture. To get However, the reproduced image data to be referred to in this case is the reproduced image data of the I picture or P picture that is one time previous in the original image data sequence with respect to the read B picture and that is one time behind. Reproduced picture data of a P picture or an I picture, and when viewed from the data series (readout series) on the recording medium, the reproduced picture data of the immediately preceding I picture or P picture and the immediately preceding P picture or I picture. It is reproduced image data. That is, the reproduced image data of P35 and P38 is referred to for B36, the reproduced image data of I32 and P35 is referred to for B33 and B34, and the reproduced image data of P29 and I32 is referred to for B30 and B31. And so on for each GOP
The B picture relating to the above will be reproduced.

By the way, in the reproducing method of the present embodiment, the next GOP is input in parallel with the decoding / reproduction of the B picture and the output of the reproduced image data of each picture, and the continuous reproduction is performed. Therefore, the switch circuits 8 and 9 are set to the connection state on the b side when the B picture of the data memory 7 is read, but are connected to the a side when the I picture and the P picture of the next GOP are input, respectively. When a picture is input, the switch circuit 8 is switched to the connection state on the b side and the switch circuit 9 is switched to the connection state on the a side. Further, as described above, the switch circuit 5 causes the I picture and P after decoding / reproduction to be performed.
It is set to the ON state only when the reproduced image data of the picture is written in the frame memory 2a, but the switch circuit 6 is set to B
When the reproduced image data of the picture is output, the connection state is switched to the a side, and when the reproduced image data of the I picture and P picture is read from the frame memory 2a and output, the connection state is switched to the b side.

In this embodiment, the switching control of the switch circuits 5, 6, 8 and 9 makes it possible to use the data memory 7 and the frame memory 2a by the FILO method. That is, in the FILO method, the B picture that has been decoded and reproduced in the data memory 7 is rewritten to the B picture of the next GOP, and the reproduced image data of the referenced I picture and P picture in the frame memory 2a is transferred to the next GOP. Related I
It is rewritten to the reproduced image data of the picture and the P picture.

Specifically, in the data input order at the time of the reverse reproduction, the data is rewritten in the data memory 7 in the usage state shown in FIG. 4 and in the frame memory 2a in the usage state shown in FIG. In each figure, the usage status of each memory 7, 2a is shown to shift from the left block to the right block in terms of time, the dotted line with an arrow indicates the data writing order, and the solid line with an arrow indicates the data. In the reading order, the two-dot chain line with an arrow indicates the data rewriting mode.

As is clear from each figure, the data memory 7
The data stored in the frame memory 2a is sequentially rewritten from the last stored data to the input data of the next GOP. Refer to the reproduced image data relating to P38 and P35 of the frame memory 2a. B36 of the data memory 7 decoded and reproduced by the input sequence [B15 of the next GOP]
→ B16 → B18 → ・ ・ ・ → B28] is rewritten to B22, and when decoding / playback is completed, reference is no longer required. The playback image data of P38 in the frame memory 2a is the input sequence [GOP] for the next GOP. I17 → P20 → P23 → ... → P29] is rewritten to P26. Further, in the next stage, B34 and B33 of the data memory 7 decoded and reproduced by referring to the reproduced image data related to P35 and I32 of the frame memory 2a are rewritten to B24 and B25 in the input sequence related to the next GOP. And the decryption
Frame memory 2a that does not need to be referenced when playback is complete
The reproduced image data of P35 in the above is rewritten to P29 in the input sequence relating to the next GOP. Then, B36, B34, and B33 which are the rewriting targets in the data memory 7 are reproduced and output through the switch circuit 6 which is in the connection state on the a side after decoding, and the P35 which is the rewriting target in the frame memory 2a has already been decoded. Since the image data is reproduced image data, it is output as it is by switching the switch circuit 6 to the b side. Note that P38 of the frame memory 2a is decoded for reference, but is not output because it is a picture temporally after the forward reproduction picture B37 at the time of the backward reproduction instruction.

Similarly, (B3 of the data memory 7)
1, B30), (B28, B27), (B25, B24), (B22, B21), ... Reproduced image data (I32, P29) in the frame memory 2a,
(P29, P26), (P26, P23), (P23, P20) ... are sequentially decoded and reproduced, but each time they are output from the data memory 7, the B picture of the next GOP (B27 , B28), (B0, B
1), (B3, B4), (B6, B7), etc., and the reproduced image data on the side of the frame memory 2a is sequentially regenerated one by one each time the reproduced image data of two B pictures is output. The reproduced image data becomes referred to, and the referred reproduced image data is output in the order of I32 → P29 → P26 → P23 → ..., and the output memory area is I2, P5,
It is rewritten to P8, P11, ....

As a result, the encoded moving image data input in the reverse reproduction direction is B36 → P35 → B34 as shown in FIG.
→ B33 → I32 → B31 → B30 → P29 → B28 → B27 → P26 →
B25 → B24 → P23 → B22 → B21 → P20 → B19 → B18 → I
17->B16->->P5->B4->B3->I2->B1-> B0 in the order of decoding / reproduction data and reproduced and output, and a moving image in the reverse order of the original image data is obtained.

According to the reproducing method of this embodiment,
As shown in FIG. 4, the frame memory 2a only needs to have a storage capacity for storing the reproduced image data for 6 frames, and a frame memory for storing the reproduced image data for 1 GOP like a conventional reproducing apparatus is required. The storage capacity can be significantly reduced compared to when it is used. On the other hand, in the reproducing method of this embodiment, the switch circuits 8 and 9 and the B before decoding / reproducing are used.
Although a data memory 7 for storing 10 pictures is required, the switch circuits 8 and 9 have a simple circuit configuration and can be easily controlled, and the data amount of the B picture before decoding / reproducing is the data amount after reproducing. Data memory 7 because it is much smaller
A small storage capacity is enough.

Further, in the present embodiment, since the encoded moving image data of Video-CD is the target of reverse reproduction, 1 GO
P is composed of 15 pictures, but 1
For encoded moving image data with a larger or smaller number of GOP pictures, the same reverse playback algorithm can be used by increasing or decreasing the storage capacity of the data memory 7 and the frame memory 2a according to the number of pictures. Of course you can.

[0030]

EFFECTS OF THE INVENTION The reverse reproduction method for compressed moving image data according to the present invention has the following effects due to the above configuration. Regarding the backward reproduction method of the compressed moving image data of the MPEG system, in the conventional method, 1 GOP worth of reproduction data is taken into consideration in consideration of the backward reproduction although the frame memory for forward reproduction is used for only two frames. Although it was necessary to prepare a frame memory that can store the data, according to the present invention, it is possible to continuously perform accurate reverse reproduction by using a frame memory having a storage capacity less than half of that. To reduce manufacturing costs.

[Brief description of drawings]

FIG. 1 is a “reverse direction reproducing method of compressed moving image data” of the present invention.
3 is a functional block circuit diagram of the reproducing apparatus in the example of FIG.

[Fig. 2] Reproduced compressed moving image data (MPEG data)
It is a figure which shows the arrangement | positioning aspect on the recording medium of this. However, the number attached to each picture indicates the order in the data sequence of the original image, and is output in the reverse order in reverse reproduction.

FIG. 3 is a diagram showing a reproduction procedure of each picture at the time of reverse reproduction.

FIG. 4 is a diagram showing a usage state of a data memory during reverse reproduction.

FIG. 5 is a diagram showing a usage state of a frame memory during backward reproduction.

FIG. 6 is a data sequence of an original image in the MPEG system
It is a figure which shows (A) and the data series (B) on a recording medium.

FIG. 7 is a functional block circuit diagram of a conventional playback device.

FIG. 8 is a diagram showing a reproduction procedure of each picture at the time of forward reproduction in a conventional reproduction device.

[Explanation of symbols]

1 ... Decoder (decoding means), 2, 2a ... Frame memory (first storage means), 3 ... Motion compensation circuit (motion compensation means), 4 ... Adder (motion compensation means), 5, 6, 8, 9 ... switch circuit, 7 ... data memory (second storage means), I ... I picture, B ... B picture,
P ... P picture.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical indication H04N 7/137 Z

Claims (1)

[Claims] 1. Variable length decoding of intra-frame coded data (I picture) which is compressed moving image data, bi-directionally predicted inter-frame coded data (B picture) and forward-predicted inter-frame coded data (P picture). Decoding means for performing dequantization / inverse quantization / inverse discrete cosine transformation, I picture reproduction data obtained by decoding I pictures by the decoding means, and P motion pictures being decoded by the decoding means and then motion compensation means to be described later. First to store processed P-picture playback data
The storage means and the I picture stored in the first storage means by reproducing the P picture decoded data decoded by the decoding means, the I picture immediately preceding the P picture in time or the P picture. Motion compensation processing is performed using picture reproduction data or P picture reproduction data, and the B picture decoded data decoded by the decoding means is an I picture that is one before and one time after the B picture. In a reproducing apparatus for compressed moving image data, which is provided with a motion compensating means for reproducing a P picture and performing a motion compensating process using the I picture reproducing data or the P picture reproducing data stored in the first storage means, If a second storage unit is provided before the decoding unit and the compressed moving image data is reproduced in the reverse direction, 1 G is stored in the first storage unit.
FILO (First) of the I picture and P picture reproduction data in the OP (Group of Picture) and the I picture reproduction data of the immediately preceding GOP in the forward direction, and the B picture in one GOP in the second storage means. -in Last-out) method, the B picture read from the second storage means is decoded by the decoding means, and the B picture after the decoding is stored.
The reproduced data of the I picture or P picture to be referred to with respect to the picture decoded data is read from the first storage means to the motion compensation means, the B picture decoded data is subjected to motion compensation processing and reproduced, and reproduced. A reverse reproduction method of compressed moving image data, characterized in that the reproduction data is output in a reverse reproduction sequence.