JP2000299853A - Device and method for decoding video - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a video decoding device capable of immediately decoding an MPEG stream by estimating a sequence header on the basis of information appearing in a picture absolutely and decoding video data by using the estimated information when the sequence header can not be detected. SOLUTION: A sequence header estimation circuit 3 uses information absolutely included in each picture and estimates information to be sent by a sequence header. When an MPEG decoder 2 can not detect the sequence header, a surface size and an aspect ratio presumed by the circuit 3 are sent to a display control circuit 4, and a display picture is set in accordance with the surface size and aspect ration estimated by the circuit 3. Thus, the circuit 3 is provided and decoding processing is performed by using the estimated surface size and aspect ratio. For this reason, reproduction can be started almost without waiting time.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention
The present invention relates to a digital video signal decoding device and method suitable for decoding a (Moving Picture Experts Group) type bit stream.

[0002]

2. Description of the Related Art Conventionally, when a digital video signal stream is decoded from an MPEG stream, first, a process of detecting a sequence header (Sequence_Header) is performed in an analysis unit in the decoding device.
This is because the image size and the aspect ratio cannot be specified without detecting the sequence header.

That is, in an MPEG stream, SD
TV (Standard Definition Television) screen and HD
A screen of TV (High Definition Television) may be sent. For example, in a digital satellite broadcast, a program of an SDTV screen and an HDTV are used in one channel.
May be multiplexed and broadcast. In some cases, the channel of the program on the SDTV screen and the channel of the program on the HDTV screen are switched.

When decoding an MPEG stream,
First, it is necessary to set an image size and an aspect ratio.
In the MPEG system, a sequence layer is defined as the highest hierarchy. In one sequence, the image size and the picture rate are the same. Then, at the beginning of each sequence, a sequence header is sent. This sequence header describes an image size, an aspect ratio, a picture rate, and the like.

For this reason, conventionally, when decoding a bit stream of the MPEG system, first, a sequence header is detected in order to set an image size and an aspect ratio. Then, after the picture size and the aspect ratio are set from the sequence header, the decoding is started from the picture coded in the first field or in the frame.

[0006]

As described above, conventionally, when decoding an MPEG stream, first, a sequence header is detected, and an image size and an aspect ratio are set. However, when decoding the MPEG stream after detecting the sequence header, there is a problem that it takes time to detect the sequence header, and a considerable waiting time is required until the reproduction is started.

That is, the sequence layer of the MPEG system is a stream having the same picture size and picture rate. The sequence header has a minimum GOP (Gropu Of
Picture) cycle, but the cycle of the sequence header is not determined. Therefore, the length of the sequence extends at most to one video program. For this reason, conventionally, for example, when a satellite broadcast channel is switched, it takes a long time to detect a sequence header, and a considerable waiting time may be required before reproduction is started.

Accordingly, an object of the present invention is to provide an MPEG
An object of the present invention is to provide a video decoding apparatus and method capable of immediately decoding an MPEG stream by estimating sequence header information and starting decoding when sequence header information of a stream is not detected.

[0009]

SUMMARY OF THE INVENTION The present invention relates to a video decoding apparatus for decoding a video stream having a hierarchical structure including a sequence layer, a GOP layer, a picture layer, a slice layer, a macroblock layer, and a block layer. A sequence header information estimating means for estimating sequence header information based on information which always appears in the data, and a decoding means for decoding video data using the information estimated by the header information estimating means when a sequence header is not detected. This is a video decoding device.

In the present invention, the sequence header information estimating means includes a vertical pixel number estimating means for estimating the number of pixels in the vertical direction of the screen from the information in the slice.

In the present invention, the sequence header information estimating means includes a horizontal pixel number estimating means for estimating the number of pixels in the horizontal direction of the screen from information in the macroblock.

In the present invention, the sequence header information estimating means estimates the number of pixels in the vertical direction of the screen from the information in the slice, and estimates the number of pixels in the horizontal direction of the screen from the information in the macroblock. An aspect ratio estimating means for estimating the aspect ratio of the screen from the number of pixels in the vertical direction and the number of pixels in the horizontal direction is included.

A slice start code (Slice_Start)
_Code) indicates the vertical position of the slice. Therefore, the slice start code (Slice_
If the fourth byte of (Start_Code) is detected, the number of pixels in the vertical direction of the screen is obtained. Also, macroblock address increment (Macroblock_Address_Increment)
) Indicates skip information of a macroblock.
Therefore, if macro block address increments are accumulated each time a macro block is decoded, the number of macro blocks in the horizontal direction of the screen can be obtained. By multiplying this by the size of the macroblock, the number of pixels in the horizontal direction of the screen can be obtained.

Using the information estimated in this way, M
If the PEG stream is decoded, the MPEG stream can be decoded immediately without detecting the sequence header.

[0015]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows an example of an MPEG decoding apparatus to which the present invention is applied. FIG.
At the input terminal 1, an MPEG (MPEG1 or M
PEG2) stream is provided. This MPEG stream is supplied to the MPEG decoder 2 and also to the sequence header estimating circuit 3. The MPEG decoder 2 decodes an MPEG (MPEG1 or MPEG2) stream.

The MPEG decoder 2 decodes the MPEG stream from the input terminal 1 and outputs a digital video data stream from the MPEG decoder 2. The output of the MPEG decoder 2 is supplied to the display 5 via the display control circuit 4. As a result, a screen based on the MPEG stream from the input terminal 1 is displayed on the display 5.

When decoding an MPEG stream by the MPEG decoder 2, it is necessary to first set an image size, an aspect ratio, and the like. These pieces of information can be detected by a sequence header (Sequence_Header). However, it sometimes takes time to detect the sequence header.

Therefore, a sequence header estimating circuit 3 is provided. The sequence header estimation circuit 3 estimates information to be sent in the sequence header by using information always included in each picture.

That is, in the sequence header, information such as an image size, an aspect ratio, a frame rate, a VBV buffer size, and a quantization matrix are transmitted. The sequence header estimation circuit 3 estimates the vertical image size using the slice information. Using the information of the macroblock, the image size in the horizontal direction is estimated. The aspect ratio is estimated using the estimated image size in the vertical direction and the image size in the horizontal direction. The frame rate is estimated by knowing the decoding timing based on a time stamp such as DTS / PTS. The output of the sequence header estimating circuit 3 is supplied to the MPEG decoder 2 and also to the display control circuit 4.

When the MPEG decoder 2 decodes an MPEG stream, if a sequence header can be detected, the information of the sequence header is used to determine the image size, aspect ratio, frame rate, VBV buffer size, quantization matrix, progressive matrix, and so on. A sequence and the like are set.

When the sequence header cannot be detected by the MPEG decoder 2, the picture size, aspect ratio, frame rate,
A VBV buffer size, a quantization matrix, a progressive sequence, and the like are set.

When the sequence header cannot be detected by the MPEG decoder 2, the image size and aspect ratio estimated by the sequence header estimating circuit 3 are sent to the display control circuit 4, and the image size estimated by the sequence header estimating circuit 3 is obtained. The display screen is set according to the aspect ratio.

As described above, the MPEG decoding apparatus according to the present invention to which the present invention is applied is provided with the sequence header estimating circuit 3 for estimating information of a sequence header. The decoding process is performed using the image size and the aspect ratio estimated by the header estimation circuit 3. For this reason, for example, even when a channel is switched by satellite broadcasting, reproduction can be started with little waiting time.

Next, the above-mentioned sequence header estimating circuit 3
The principle and configuration of will be described specifically.

As shown in FIG. 2, the MPEG data structure has a hierarchical structure of a sequence layer, a GOP layer, a picture layer, a slice layer, a macro block layer, and a block layer.

The sequence layer is a group of screens having a series of the same attributes, for example, the same image size and the same image rate. The GOP layer is a group of screens serving as random access units. The picture layer is a group of attributes common to one screen. The slice layer is a group obtained by subdividing one screen. The macroblock layer is a group obtained by further subdividing the slice layer, and is a group for performing motion vector detection and the like. Block layer is DC
It is the block itself for performing T conversion.

At the head of the sequence, a sequence header (SH: Sequence_Header) is provided. In this sequence header, (1) Horizontal_Size_Value: the number of pixels in the horizontal direction of the image (2) Vertical_Size_Value: the number of pixels in the vertical direction of the image (3) Aapect_Ratio_Information: aspect ratio (4) Frame_Rate_Code: display cycle of the image (5) VBV_Buffer_Size: Size of generated code amount control virtual buffer (VBV) (6) Load_Quantization_Matrix: Quantization matrix for macroblock (7) Progressive_Sequence: Indicates that it is sequential scanning, etc.

Of these, the number of pixels in the vertical direction of the image (Vert
ical_Size_Value) can be estimated from the slice information.

FIG. 3 shows the structure of a slice in one picture. As shown in FIG. 3, one picture is divided into a plurality of slices. The leftmost slice of the top row (Bs = 1) in one picture is Slice.
(1, 0), and the second slice from the left at the top is Slice (1, 1). Hereinafter, similarly, the leftmost slice of the bottom row (Bs = N) in one picture is Sl
ice (N, 0), and the second slice from the left at the bottom is Slice (N, 1).

As described above, a plurality of slices can exist in the horizontal direction, but the slice is always switched to a slice having a new slice ID as a start code at the left end of the picture. From this, if a slice is used, the vertical size of one picture can be estimated.

That is, a slice start code (Slice_Start_Code), which is a synchronization code indicating the start of a slice layer, is inserted at the beginning of each slice. This slice start code is represented by hexadecimal "00 00
01 01 to AF ”, and the last 1 of the code.
The byte (“01 to AF” in the fourth byte) indicates the vertical position of the slice in hexadecimal. Thus, the fourth byte of the slice start code (Slice_Start_Code) corresponds to the vertical position of the slice, which is the same number in the same line.

From the above, if the fourth byte of the slice start code at the bottom of one picture is detected,
The number of pixels in the vertical direction of the screen can be estimated.

FIG. 4 is a functional block diagram showing a configuration for estimating the number of pixels in the vertical direction of the screen from slice information as described above. In FIG. 4, an MPEG stream is supplied to an input terminal 11. Picture start code (Picture_Start_) in this MPEG stream
Code) is detected by the picture start code detection unit 12, and the slice start code (Slice_Start_Code) is detected.
Is detected by the slice start code detection unit 13.

Picture start code (Picture_Star)
t_Code) is a code indicating the start of the picture layer.
The slice start code (Slice_Start_Code) is
In the code indicating the start of the slice layer, the fourth byte of the slice start code indicates the vertical position.

The output of the slice start code detector 13 is sent to the fourth byte extractor 14. The fourth byte extracting unit 14 extracts the information of the fourth byte of the slice start code. The output of the fourth byte extracting unit 14 is sent to the vertical size register 15.

When the picture start code is detected by the picture start code detecting section 12, the head of one picture can be detected. When the head of one picture is detected, the vertical size register 15 is reset. Then, it is determined whether the slice start code is detected by the slice start code detection unit 13.

When the slice start code is detected by the slice start code detector 13, the fourth byte information of the slice start code is extracted by the fourth byte extractor 14, and the information of the fourth byte of the slice start code is stored in the register. 15 is supplied. When the slice start code is detected by the slice start code detection unit 13 until the next picture start code is detected by the picture start code detection unit 12, the fourth byte extraction unit 14 outputs the fourth byte of the slice start code. The information is extracted and the register 15 is updated with this value.

The picture start code detecting section 1
When the next picture start code is detected in step 2, the information in the register 15 is taken into the vertical size register 16 as information on the number of pixels in the vertical direction.

As described above, the slice start code detection unit 13 detects the slice start code from the beginning to the end of the picture, and when the slice start code is detected, the fourth byte extraction unit 14 detects the fourth byte of the slice start code. If a value is extracted and taken into the vertical size register 15, an estimated value of the number of pixels in the vertical direction can be obtained from the value of the vertical size register 16.

The number of pixels in the horizontal direction of the image (Horizontal_
Size_Value) can be estimated from macroblock information. That is, as shown in FIG. 5, a macroblock is obtained by further dividing a slice. In this example, slice Slice (1, 0) includes macro blocks MB1, MB2, MB3, and slice Sli
There are MB4, MB5, MB6 and MB7 in ce (1,1). A skipped macroblock exists between the macroblocks MB2 and MB3 of the slice Slice (1, 0).

At the beginning of a macroblock, a macroblock address increment (Macroblock_Address_In)
crement). This macroblock address increment is a VLC (variable length code) indicating a macroblock to be skipped, and is “1” in a normal adjacent macroblock, but is skipped when a skipped macroblock exists. The value is increased by the number of macro blocks.

Accordingly, the value of the macroblock address increment within one slice is accumulated for each macroblock. When there are a plurality of slices in the horizontal direction, the increment of the macroblock address increment accumulated in each slice is obtained. By adding the values, the number of macroblocks in the horizontal direction per screen can be recognized. If this is multiplied by the size of the macroblock, ie, “16” for the luminance signal and “8” for the two color difference signals, the number of pixels in the horizontal direction per picture can be estimated.

FIG. 6 is a functional block diagram showing a configuration for estimating the number of pixels in the horizontal direction of the screen from the information of the macro block as described above.

In FIG. 6, an MPEG stream is supplied to an input terminal 21. The slice start code (Slice_Start_Code) in this MPEG stream is detected by the slice start code detection unit 22, and the macroblock address is incremented (Macroblock_Address_).
Increment) is detected by the macroblock address increment detection unit 23.

When estimating the number of pixels in the horizontal direction, the first picture coding type (Picture Codi
Regardless of what ng Type), the MPEG decoder 2 (FIG. 1) decodes only that picture by intra processing. When the macroblock is decoded, a signal indicating that the macroblock has been decoded is output. A signal indicating that the macroblock has been decoded is supplied from an input terminal 22 to a decoded macroblock detector 27.

When the slice code detector 22 detects the leftmost slice of one line, the register 25
Registers 28A, 28B,..., Horizontal size register 31
Is cleared. Then, the macroblock address increment detection unit 23 detects a macroblock address increment (Macroblock_Address_Increment).

This macro block address increment indicates skip information of the macro block, and is "1" to "1".
The increment value corresponding to “33” is written in a variable length code. When the macroblock address increment is “33” or more, macroblock escape (Macroblock_Escape) is also referred to.

The output of the macro block address increment detection section 23 is supplied to the VLC decoding section 24. VL
The C decoding unit 24 decodes the value of the macroblock address increment.

The output of the VLC decoder 24 is supplied to an adder 26. The output of the register 25 is supplied to the adder 26. The register 25 is supplied with the output of the decode macroblock detection unit 27. When it is detected from the output of the decoded macroblock detection section 27 that the decoding of the macroblock has been performed, the adder 26 calculates the value of the current macroblock address increment and the value of the previous macroblock address increment. The values are added, and the value of the macroblock address increment is accumulated. As a result, the accumulated value of the macroblock address increment in each slice of the same horizontal line is obtained.

The output of the register 25 is the register 28A,
8B,... Registers 28A, 28B, ...
When there are a plurality of slices in the horizontal direction, the stored value of the macroblock address increment in each slice is fetched. The output of the slice start code detection unit 22 is supplied to the registers 28A, 28B,. By the output of the slice start code detection unit 22, the accumulated value of the macro block address increment is taken into the registers 28A, 28B,. For example, the accumulated value of the macroblock address increment in the first slice of the same horizontal line is taken into the register 28A, and the accumulated value of the macroblock address increment in the next slice of the same horizontal line is taken into the register 28B. Go.

The outputs of the registers 28A, 28B,... Are supplied to the adder 29. In the adder 29, the accumulated values of the macroblock address increment in each slice are added. As described above, the number of macroblocks in the horizontal direction per screen is obtained by summing up the macroblock address increment values accumulated in each slice.

The output of the adder 29 is supplied to the multiplier 30. The multiplier 30 multiplies the number of macroblocks by the size of the macroblock to calculate the number of pixels in the horizontal direction. That is, the multiplier 30 calculates the number of pixels in the horizontal direction per picture by multiplying the number of macroblocks by the horizontal size of the macroblock. The number of pixels in the horizontal direction obtained in this manner is supplied to the horizontal size register 31.

As described above, the adder 26 and the register 25
Then, the macroblock address increment value in one slice is accumulated for each macroblock, and the adder 29 adds up the macroblock address increment values accumulated in each slice, so that a horizontal picture per screen is obtained. The number of macroblocks in the direction is calculated. In a multiplier 30, this is multiplied by the size of the macroblock, and
The number of horizontal pixels per picture is determined.

Aspect ratio (Aapect_Ratio_Inform)
ation) can be estimated from the number of pixels in the horizontal direction and the number of pixels in the vertical direction of the image obtained as described above. If the image size is (720 × 480), since it is SDTV, the aspect ratio is estimated to be (4: 3). If the image size is (920 × 1080), it is HDTV, so (1
6: 9).

The display cycle of an image (Frame_Rate_Code) is indirectly inferred by knowing the decoding timing based on a time stamp such as DTS / PTS.

The generated code amount control virtual buffer (V
As for the size (VBV_Buffer_Size) of the (BV), the largest possible level profile is prepared. Similarly, for a decoded image, a capacity of the maximum size is secured in a normally considered applicable level profile.

The quantization matrix for the macro block (Lo
For ad_Quantization_Matrix), a default value is used instead.

The progressive sequence (Progressive__)
Sequence) is replaced by a progressive frame (Progressive_Frame) in a picture coding type (Picture_Coding_Type) that is multiplexed every frame.

In the above example, MPEG1 or MP1
Although the case where the stream of the EG2 system is decoded has been described, the present invention is similarly applicable to the case of decoding a stream having a similar hierarchical structure.

[0060]

According to the present invention, when the sequence header is not detected, the slice start code (Slic
e_Start_Code), the number of pixels in the vertical direction of the screen is estimated, the macroblock address increment (Macroblock_Address_Increment) is accumulated, and the number of macroblocks in the horizontal direction of the screen is obtained.
This is multiplied by the size of the macroblock to estimate the number of pixels in the horizontal direction of the screen. Then, MPEG decoding is performed using the information estimated in this manner. Thus, the MPEG stream can be immediately decoded without detecting the sequence header.

[Brief description of the drawings]

FIG. 1 is a block diagram of an example of an MPEG decoding device to which the present invention has been applied.

FIG. 2 is a schematic diagram used for explaining a hierarchical structure of the MPEG system.

FIG. 3 is a schematic diagram used for explaining a slice.

FIG. 4 is a functional block diagram illustrating a configuration of a circuit for estimating a horizontal image size.

FIG. 5 is a schematic diagram used for describing a macroblock.

FIG. 6 is a functional block diagram showing a configuration of a vertical image size estimation circuit.

[Explanation of symbols]

2 ... MPEG decoder, 3 ... Header estimation circuit,
4 ... display control circuit, 13 ... slice start code detection unit, 14 ... 4th byte extraction unit, 23 ...
Macro block address increment detector

Claims (8)

[Claims] 1. A video decoding apparatus for decoding a video stream having a hierarchical structure including a sequence layer, a GOP layer, a picture layer, a slice layer, a macroblock layer, and a block layer. A video decoding apparatus comprising: a sequence header information estimating unit for estimating header information; and a decoding unit for decoding video data using the information estimated by the header information estimating unit when the sequence header is not detected. . 2. The sequence header information estimating means,
2. The video decoding apparatus according to claim 1, further comprising a vertical pixel number estimating unit for estimating the number of pixels in the vertical direction of the screen from information on the slice. 3. The sequence header information estimating means includes:
2. The video decoding apparatus according to claim 1, further comprising a horizontal pixel number estimating unit for estimating the number of pixels in the horizontal direction of the screen from information in the macroblock. 4. The sequence header information estimating means includes:
The number of pixels in the vertical direction of the screen is estimated from the information in the slice, the number of pixels in the horizontal direction of the screen is estimated from the information in the macroblock, and the number of pixels in the vertical direction and the number of pixels in the horizontal direction are estimated from the estimated number of pixels in the horizontal direction. 2. The video decoding apparatus according to claim 1, further comprising aspect ratio estimating means for estimating an aspect ratio of a screen. 5. A video decoding method for decoding a video stream having a hierarchical structure including a sequence layer, a GOP layer, a picture layer, a slice layer, a macroblock layer, and a block layer, wherein the sequence is based on information that always appears in a picture. A video decoding method for estimating header information and decoding video data using the estimated information when the sequence header is not detected. 6. The video decoding method according to claim 5, wherein the estimation of the sequence header information is performed by estimating the number of pixels in a vertical direction of a screen from information in a slice. 7. The video decoding method according to claim 5, wherein the estimation of the sequence header information is performed by estimating the number of pixels in a horizontal direction of a screen from information in a macroblock. 8. The estimation of the sequence header information includes estimating the number of pixels in a vertical direction of a screen from information in a slice,
6. The screen aspect ratio according to claim 5, wherein the number of pixels in the horizontal direction of the screen is estimated from information in the macroblock, and the aspect ratio of the screen is estimated from the estimated number of pixels in the vertical direction and the number of pixels in the horizontal direction. Video decoding device.