JP3341842B2 - Buffer management method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a buffer management method, and more particularly to a variable length coded data suitable for controlling a generated code amount and joining or reading out a plurality of variable length coded data. A buffer management method for generating a buffer.

[0002]

2. Description of the Related Art MPEG (Moving Picture Experts Group) is known as a typical example of variable length coding of information. In this MPEG, an encoded bit stream has a variable length code amount for each picture in the case of a video signal. This is because MPEG uses discrete cosine transform (DC
T), quantization and Huffman coding are used at the same time as the information conversion, and at the same time, it is necessary to adaptively change the code amount allocated to each picture in order to improve image quality, and motion compensation prediction is performed. Therefore, the entropy of the coded image itself greatly changes, for example, in some cases, the input image is encoded as it is, and in some cases, the difference image which is the difference between the predicted images is encoded. In this case, in many cases, the code amount is controlled while allocating to the entropy ratio of the image and keeping the buffer limit.

FIG. 5 is a block diagram showing an example of a conventional buffer control device. In the figure, input image data is input to a variable-length encoder 11 and subjected to variable-length encoding, and then input to a buffer manager 12. The buffer manager 12 monitors the relationship between the input code amount and the coding rate, sets a target code amount so as to be contained in a predetermined buffer, and feeds back to the variable length encoder 11. The target code amount is input to a code amount controller (not shown), and the generated code amount is suppressed by increasing the quantization value set in the quantizer, or the generated code amount is increased by decreasing the quantization value. .

When such variable-length coded data is coded at a fixed transfer rate (coding rate), data is input at a constant speed when the buffer amount of the decoder is set to a maximum value as shown in FIG. Then, from the place where only a predetermined value has accumulated,
A predetermined time unit (1 / NTSC video signal)
It is specified by MPEG that a virtual decoder model that performs decoding instantaneously in units of 29.97 seconds is connected to the output of the encoder and used to encode the buffer so that neither overflow nor underflow occurs. ing.

In FIG. 1, I, P and B are I pictures (intra-frame coded pictures), P pictures (inter-frame forward coded pictures) and B pictures (bi-directional coded pictures) specified by MPEG. ) Indicates that these are stored in the buffer, the decoding of each picture is performed instantaneously every 1 / 29.97 second, and data is instantaneously extracted from the buffer.

[0006] This is called VBV (Video Buffering Verifie).
r), the details of which are described in the International Organization for Standardization (I
SO) according to ISO-11172-2, ISO1381
8-2. If you follow this rule, V
Although the rate in the BV buffer changes locally, if the observation time is long, the transfer rate becomes fixed, and M
PEG defines this as a fixed rate.

In the case of the fixed transfer rate, the code amount changes as shown in FIG. 6 in a steady state. When the generated code amount is small, the transition state is stuck to the upper limit as shown in FIG. In FIG. 7, in the picture indicated by the four arrows, even if the generated code amount is extremely small, the code amount must be increased by adding an invalid bit so as not to overflow.

In the case of the variable transfer rate, the above definition of the fixed transfer rate is extended, and when the buffer occupancy value reaches the upper limit value, the reading of the decoder is stopped, so that overflow does not occur in principle. Is defined as In the code amount change diagram shown in FIG. 8, the picture indicated by the four arrows has a very small amount of generated code, but since reading of the decoder is stopped, invalid bits need to be inserted as in the case of a fixed transfer rate. There is no. Therefore, encoding is performed so that only underflow does not occur.

[0009]

However, in the conventional variable length coded data processing method and buffer control apparatus, when the coded video data is edited and combined in units of, for example, 45 frames, or when several video compression When the read destination is changed to another video stream in units of, for example, 45 frames, an overflow or an underflow may occur at a fixed transfer rate, and an underflow may occur at a variable transfer rate.

This will be described with reference to FIG.
In the case of the fixed transfer rate, the temporal change of the VBV buffer occupancy value of the video data is represented by three video data to as shown in FIG. 9A, and at the connection point 1 the video data and the video data are connected. At the connection point 2, video data and video data are connected, but are normally connected without overflow or underflow.

The temporal change of the VBV buffer occupancy value of the video data is as shown in FIG. 9B. At connection point 1 indicated by, video data and video data are connected, and at connection point 2, video data is connected. The data and the video data are connected, but are normally connected without overflow or underflow.

However, as shown in FIG. 9C, when video data is connected at the connection point 1 (the 45th frame), the buffer occupancy value of the first picture of the video data is changed to the end of the video data. Since it is necessary to set the buffer occupation value at the time, the video data is shifted upward as a whole as compared with FIG. 9B, and an overflow occurs at the 47th frame and the like.

When video data is connected, the buffer occupancy value of the first picture of the video data is set to the buffer occupancy value at the end of the video data and the connection is established. Occurs.

On the other hand, the temporal change of the VBV buffer occupancy value of the video data in the case of the variable transfer rate is as shown in FIG. 9D, for example. That is, when video data and video data corresponding to the video data are connected at the connection point 1 (the 45th frame), the buffer occupancy value of the first picture of the video data is equal to or less than the buffer occupancy value at the end of the video data. However, in the video data of the variable transfer rate, the overflow of the decoder is stopped at the upper limit of the buffer, and the overflow does not occur.

However, when video data is connected, the buffer occupancy value of the first picture of the video data is set to be equal to or less than the buffer occupancy value at the end of the video data. As shown, an underflow occurs in the 90th frame and thereafter as in the case of the fixed transfer rate.

Here, "underflow" means that data has not yet arrived at the time when the picture was decoded by the reproducer. Therefore, in this case, although depending on how the reproducing device is made, generally, the immediately preceding picture is reproduced again (freeze), and the process waits until all the data are collected.

On the other hand, "overflow" means that the data corresponding to the overflow of the buffer of the regenerator is lost. Generally, MPE
G employs Huffman coding, and if some data is lost, the entire picture may be lost. In addition, since the motion compensation prediction is performed, an image predicted with reference to the picture is also lost, and there is a possibility that not only an image but also a temporal direction may be lost.

Further, the limitation of the buffer is that the connection is made at a predetermined connection point with the same buffer occupancy value at the fixed transfer rate, and at the variable transfer rate with a value that is assumed to be less than the buffer occupancy value of the previous picture. If started, there is no restriction on the buffer transition within the predetermined picture unit. Therefore, in the case of simple operation, the code amount of the last picture in the predetermined picture unit is strongly controlled to achieve the target code amount. Will be encoded.

However, in the case of a fixed transfer rate, etc., 1
It is necessary to match with bit precision, and when encoding in real time, it is impractical. Also, even in the case of a variable transfer rate, the connection point must be controlled so as to be equal to or larger than the buffer occupancy value of the previous picture, and the code amount of the last picture in a predetermined picture unit is limited to a considerably small direction. Therefore, there is a risk that the image quality periodically deteriorates.

The present invention has been made in view of the above points,
It is an object of the present invention to provide a buffer management method capable of preventing data loss due to buffer overflow of a regenerator and minimizing adverse effects due to underflow.

[0021]

In order to achieve the above object, the present invention provides the following method. (1) When obtaining compressed data by variable length coding,
A decoder buffer for storing the compressed data when decoding the compressed data, assuming a decoder buffer having a lower limit value and an upper limit value in the occupation value of the compressed data to be stored; The target code amount for controlling the code amount to be generated for each unit of generation of the compressed data is controlled so as not to fall below the value, and the generated code amount is controlled. In the buffer management method of monitoring by generation unit,
The target code amount in the one generation unit so that the occupied value of the buffer at the end of the generation of the compressed data of one generation unit is the first buffer value set between the lower limit value and the upper limit value. And sets the occupancy value of the buffer at the start of generation of the next generation unit of compressed data following the one generation unit of compressed data to a second buffer value equal to or smaller than the first buffer value. A buffer management method characterized by setting to a value.

By using the method of the present invention described in the above (1),
Overflow or underflow that occurred when the variable length coded data of a plurality of information was edited and connected in the generation unit of the compressed data, or when the read destination was changed in the generation unit of several compressed streams. Disappears.

[0023]

Next, an embodiment of a buffer management method of the present invention will be described with reference to the drawings. First, a first embodiment will be described. This embodiment relates to a variable transfer rate. FIG. 1 is a block diagram of a buffer control device to which the first embodiment is applied. This buffer control device includes a variable length encoder 21, a buffer value reset unit 22, a buffer manager 23, and a target code amount limiter 24.

The input image data is input to the variable-length encoder 21 and is subjected to variable-length encoding. The buffer manager 23 monitors the relationship between the code amount of the input variable-length coded data and the coding rate,
The target code amount is set so as to fall within a predetermined buffer value between the lower limit and the upper limit. This target code amount is input to the target code amount limiter 24, and in the first embodiment, a picture immediately before the connection point, that is, a predetermined picture based on the picture count information from the variable length encoder 21, The buffer value of the last picture in a picture unit (GOP) is limited to a target code amount that converges on a first buffer value set between a lower limit and an upper limit.

The target code amount limited by the target code amount limiter 24 is fed back to the variable length encoder 21 and input to a code amount controller (not shown), where the quantization value to be set in the quantizer is set. The generated code amount is suppressed by increasing the value, or the generated code amount is increased by reducing the quantization value. At the time of inputting the first picture in a predetermined picture unit, the buffer value reset unit 22 generates a reset signal based on the picture count information from the variable length encoder 21 and supplies the reset signal to the buffer manager 23. The buffer manager 23 is reset to a second buffer value equal to or smaller than the first buffer value at the beginning of the connection point based on the reset signal.

FIG. 3A shows that the target code amount is limited so that the buffer value of the last picture of the first video data I in a predetermined picture unit becomes the first buffer value V1, and This shows a state in which the buffer value of the first picture of the second video data II in a predetermined picture unit following the last picture of the video data I has been reset to the second buffer value V2 by the buffer value reset unit 22.

When the two encoded video data I and II are connected as shown in FIG. 3B. That is, the second video data II is calculated by subtracting the second buffer value V2 from the first buffer value V1.
As shown by II ', the whole is shifted upward and combined. For this reason, there is a portion where the buffer occupancy value of the second video data exceeds the upper limit value of the buffer occupancy value and tends to overflow. At this time, the reading of the decoder is stopped, and as shown by a. There is no problem because the data is read so that no overflow occurs.

Next, another embodiment (second embodiment) will be described. This embodiment is also an embodiment relating to a variable transfer rate. The block configuration of a buffer control device to which the second embodiment is applied is the same as that shown in FIG. In the second embodiment, the target code amount limiter 24 uses the buffer value V of the last picture in the picture unit immediately before the connection point based on the picture count information.
In order to converge to 1, a monotonic increase is gradually performed for each predetermined picture from a buffer lower limit value in a steady state (for example, 0) to a value slightly lower than the buffer value V1 of the last picture (for example, a value about 20% lower than V1). To generate a limit value. This state is shown in FIG. As shown in the figure, the lower limit of the first video data III in a predetermined picture unit is the first video data III.
, And is connected to the second video data IV at a connection point.

Here, the lower limit value of the buffer within a predetermined picture unit may be determined by increasing the lower limit value monotonically as shown by V in FIG. The buffer occupancy value may be monotonically increased from the picture, that is, the buffer occupancy value may gradually increase toward a value slightly lower than the first buffer value V1 within a predetermined picture unit. In this embodiment, the generated code amount is gradually limited by gradually increasing the lower limit value. Therefore, extreme image degradation or only the last picture in a predetermined picture unit is deteriorated, and the image quality is periodically reduced. Dangers such as deterioration of the body can be prevented.

The predetermined picture unit may not always be a fixed unit (here, 45 frame times).
It may be 0 frame time or 15 frame time. In this case, the gradient of the lower limit characteristic of the buffer is controlled in accordance with the reproduction time of the compressed data in the unit of generation time of the compressed data, by decreasing the gradient when the length is long and increasing the gradient when the length is short. , The most appropriate buffer control can be performed. That is, the slope of the monotonically increasing lower limit value of the buffer in the target code amount limiter 24 depends on the reproduction time of the compressed data in the generation time unit of the variable compressed data, that is, the variable time when the predetermined picture unit is variable. May be changed adaptively. As described above, when the gradient of the lower limit value characteristic of the buffer is changed according to the reproduction time of the compressed data of the generation unit, the most appropriate buffer control can be performed for the control unit.

When a target code amount that is equal to or smaller than the lower limit value is allocated, the target code amount is limited by the target code amount limiter 24. On the other hand, even if the target code amount is reduced, the buffer manager 23 generally feeds back a certain percentage of the picture to be coded later by that amount, so that it is not possible to obtain the target coding rate when viewed as a whole. Easy.

Next, still another embodiment (third embodiment) will be described. FIG. 2 is a block diagram of a buffer control device to which the third embodiment is applied. In the figure, the same components as those of FIG. 1 are denoted by the same reference numerals, and description thereof will be omitted. 2 includes a variable length encoder 21, a buffer value reset unit 22, and a buffer management unit 2.
3, a target code amount limiter 24 and an invalid bit generator 25.

This embodiment is an embodiment relating to a fixed transfer rate. The target code amount limiter 24 determines that the buffer occupation value in the VBV of the last picture in a predetermined picture unit is equal to or greater than the buffer value V1 (= V2). The target code amount is set smaller beforehand, and the difference from the target buffer value V1 (= V2) is set as invalid bit amount information.
It is supplied to the invalid bit generator 25. Invalid bit generator 2
5 inserts invalid bits after the last picture in a predetermined picture unit by the input invalid bit amount. The invalid bit is, for example, all-zero data.

V of video data in this embodiment
FIG. 3C shows an example of a temporal change of the buffer occupancy value of the BV.
Shown in As shown in FIG. 3C, the target code amount is set such that the buffer occupation value V3 of the last picture in the predetermined picture unit in VBV is equal to or larger than the buffer value V1 (= V2). Further, a difference value (V3-V1) between the buffer occupation value V3 and the target buffer value V1 (= V2) is input to the invalid bit generator 25 as the invalid bit amount SB. By being inserted after the last picture, the buffer occupancy value in the VBV of the last picture is set to the target buffer value V1 (= V2).

In this embodiment, at a fixed transfer rate, the buffer occupancy value of the last picture in a predetermined picture unit of VBV of video data is set to the target buffer value V
1 (V2) can be controlled without error, so that connection to different video data can be made so that overflow does not occur.

It should be noted that the present invention is not limited to the above embodiment, and the variable-length encoded data to be combined (connected) may be data other than video data. The present invention can be applied to any type of data as long as the data is reproduced by having a buffer in a decoder (reproducer).

[0037]

As described above, according to the present invention,
Overflow or underflow that occurred when the variable length coded data of a plurality of information was edited and connected in the generation unit of the compressed data, or when the read destination was changed in the generation unit of several compressed streams. Can be eliminated, so that the degree of freedom in editing and changing the reading destination can be increased as compared with the related art. Further, according to the present invention, by changing the target buffer value (V1 or V2), editing such as addition and insertion can be performed in a predetermined generation unit in compressed data that ends with any buffer value. Enabled as compressed data.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a buffer control device to which the first and second embodiments are applied.

FIG. 2 is a block diagram illustrating a buffer control device to which a third embodiment is applied;

FIG. 3 is a VBV for explaining the operation of the first and third embodiments.
FIG. 5 is a time transition diagram of the buffer occupancy value of FIG.

FIG. 4 is a time transition diagram of VBV buffer occupancy values for explaining the operation of the second embodiment.

FIG. 5 is a block diagram of an example of the related art.

FIG. 6 is a diagram illustrating a temporal transition of a buffer occupancy value of MPEG VBV.

FIG. 7 is a time transition diagram of an example of a buffer occupancy value of a fixed transfer rate VBV.

FIG. 8 is a time transition diagram of an example of a buffer occupancy value of a variable transfer rate VBV.

FIG. 9 is a temporal transition diagram of a VBV buffer occupancy value when editing and reading are changed by a conventional method.

[Explanation of symbols]

 Reference Signs List 21 variable length encoder 22 buffer value resetter (reset means) 23 buffer manager 24 target code amount limiter 25 invalid bit generator (insertion means) V1 first buffer value V2 second buffer value V3 third Buffer value SB Invalid bit amount

Continuation of the front page (56) References JP-A-2002-176646 (JP, A) JP-A-2-194734 (JP, A) JP-A-7-184196 (JP, A) JP-A 8-84333 (JP, A) A) JP-A-8-149408 (JP, A) JP-A-8-237599 (JP, A) JP-A-8-331560 (JP, A) JP-A-9-74557 (JP, A) JP-A-9 −93136 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H03M 7/40 H04N 7/24

Claims (1)

(57) [Claims] A decoder buffer for accumulating compressed data at the time of decoding the compressed data when obtaining the compressed data by variable length coding, wherein a lower limit value and an upper limit value of an occupation value of the compressed data to be stored are determined. And a target code amount for controlling a code amount to be generated for each compressed data generation unit so that the occupied value of the buffer does not fall below the lower limit value. In a buffer management method for controlling an amount and monitoring a transition of an occupancy value of the buffer in a unit of generation of the compressed data, the occupancy value of the buffer at the end of generation of compressed data in one generation unit includes the lower limit value and the lower limit value. The target code amount in the one generation unit is set so as to be a first buffer value set between the upper limit value and the compressed data of the one generation unit. And setting a buffer occupation value at the start of generation of compressed data of a next generation unit following the second buffer value to a second buffer value equal to or smaller than the first buffer value.