KR100488018B1 - Moving picture coding method - Google Patents

Abstract

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to video coding, and more particularly, to a reference picture calculation method in RPSL in which efficient video coding can be performed by reducing the amount of bits used in a reference picture selection layer (RPSL).

Reference picture calculation method in the RPSL according to the present invention, the field and code number used for the RMPNI (Re-Mapping of Picture Numbers Indicator) operation of the Reference Picture Selection Layer (RPSL) in a video coding system using multiple reference pictures Indicates a picture number information of a target stored in a short-term buffer, a field represented by a DELTA value having a sign, and a code number thereof; A field for indicating picture number information of a target stored in a long-term buffer and a code number thereof; And a field for indicating the end of the RMPNI operation and a code number thereof.

In addition, the reference picture calculation method in the RPSL according to the present invention, the DPN (Difference Picture Numbers) used for the MMCO (Memory Management Control Operation) operation of the Reference Picture Selection Layer (RPSL) in a video coding system using multiple reference pictures ) Field is represented by a DPN value with a sign.

Description

Moving picture coding method

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to video coding, and more particularly, to a video coding method in which efficient video coding can be performed by reducing the amount of bits used in a reference picture selection layer (RPSL).

A video coding system that performs motion compensation from multiple reference pictures, for example, multiple picture buffers of video coding systems such as Annex U, H.26L and MPEG-4 part 10 of H.263. The picture buffer consists of a short-term buffer and a long-term buffer.

Such a video coding system allows various types of indexing so that a picture to be coded can efficiently refer to a reference picture stored in two buffers. This is possible through a Re-Mapping of Picture Number Indicator (RPMNI) operation. Do. There are four types of fields used for RMPNI operation, and code numbers are mapped from 0 to 3 according to each field. At this time, each code number already has a unique codeword, and the relationship between the code number and the codeword length is as shown in [Table 1].

Code number Codeword Length (bits) 0 One 1 ~ 2 3 3 ~ 6 5 7-14 7 15-30 9 ... ...

As is known, codewords corresponding to the code numbers' 0 ',' 1 ',' 2 ',' 3 ',' 4 '... etc. are respectively' 1 ',' 010 ',' 011 ',' 00100 ',' 00101 '... etc. Accordingly, the length of the codeword corresponding to the code number '0' is 1 bit, and the length of the codeword corresponding to the code number '1 ~ 2' is 3 bits. As the number increases, the length (bit) of the corresponding codeword also increases.

In addition, as shown in [Table 2], each code number of the RMPNI performs a different operation, and code numbers '0' and '1' respectively represent absolute values of ADPN (negative and negative signs). The Absolute Difference of Picture Numbers field is used to indicate the Picture Number (PN) of the target picture, and code number '2' is used for the Long-term Picture Index for Re-Mapping (LPIR) field. The target long term picture is indicated, and finally the code number '3' indicates the end of the RMPNI operation.

RMPNI code number field 0 ADPN value with a negative sign One ADPN value with a positive sign 2 LPIR representing LPIN 3 End

Meanwhile, RMPNI converts default indexing of multiple picture buffers into relative indexing in order to increase coding efficiency of a picture to be coded. Here, the picture having the target PN calculated by the RMPNI field or the picture having the LPIR is indexed in the order of 0, 1, and 2 according to the RMPNI command order, and the other pictures are automatically pushed back by the index number.

At this time, the ADPN field is mapped to a code number having a value of 'ADPN-1', and a codeword having the code number is assigned, and the LPIR field is mapped to a code number having the same value, and a codeword of the code number is assigned. . Here, since the ADPN can never have a value of 0, the code number of the ADPN corresponds to 'ADPN-1'.

For example, if three pictures with PN = 102, 101, and 100 are stored in the short-term buffer, and two pictures with Long term Picture Index (LPIN) = 0 and 1 are stored in the long-term buffer, the default index is: Becomes the same as

Default Index 0: PN = 102

Default index 1: PN = 101

Default index 2: PN = 100

Default index 3: LPIN = 0

Default index 4: LPIN = 1

If the PN of the current picture to be coded is 103 and coded with the following index, it is assumed that the coding efficiency can be improved.

Relative Index 0: PN = 101

Relative Index 1: LPIN = 0

Relative Index 2: PN = 102

Relative Index 3: PN = 100

Relative Index 4: LPIN = 1

If the ADPN has not been indexed before, the PNP (Picture Number Prediction) becomes the PN of the current picture, that is, 103, and the target PNQ (Picture in Question) is 101. Therefore, the ADPN is defined by 'PNQ-PNP'. -2 '. In addition, the second target picture becomes a long term picture with 'LPIR = 0'. Therefore, three RMPNI commands are executed.

RMPNI ADPN = -2-> Code number 0 2 (1 bit + 3 bit = 4 bit)

RMPNI LPIR = 0-> Code No. 2 0 (3 bit + 1 bit = 4 bit)

RMPNI end-> code number 3 (5 bit)

Accordingly, all 13 bits are used for the RMPNI operation.

Meanwhile, in a video coding system using multiple reference pictures, the coded pictures are automatically stored in the short-term buffer of the multiple picture buffer so that they can be used as reference pictures. However, when it is necessary to change the storage state of reference pictures stored in the multiple picture buffer depending on the situation, such as when the picture no longer needs to be stored in the short term buffer or in the long term buffer, a reference picture buffering type ) And operations such as Memory Management Control Operation (MMCO).

Here, the RPBT is coded pictures are stored in the first-in-first-out (FIFO) structure in the short-term buffer, and if there is no change in the buffer, a value of '0' is assigned, and if the characteristic change of the stored pictures in the buffer is required Is assigned a value of '1'. When the RPBT has a value of '1', a characteristic change occurs in the short term buffer and the long term buffer, which are performed by the MMCO instruction. [Table 3] summarizes the code number and operation contents corresponding to the MMCO instruction and the following fields.

MMCO Code Number Calculation contents field 0 End - One Remove the specified short-term picture from the buffer DPN 2 Remove the specified long-term picture from the buffer LPIN 3 Move the specified short-term picture to the long-term buffer DPN, LPIN 4 Determining Long Buffer Size MLIP1 5 Reset all contents of multiple picture buffer -

In this case, DPN (Difference Picture Numbers) can be obtained by 'PNC-PNQ', and PNC (Current decoded Picture Number) indicates PN of a currently coded picture. Here, since the value of the DPN includes '0' and is always positive, the DPN is mapped to a code number having the same value and assigned a codeword having the code number.

The present invention proposes a setting of a field and a code number used for a Re-Mapping of Picture Numbers Indicator (RPMNI) operation of a reference picture selection layer (RPSL), thereby reducing the amount of bits of the RMPNI operation, thereby performing efficient video coding. The purpose of this is to provide a video coding method that allows.

In addition, the present invention proposes to set the DPN (Difference Picture Numbers) field and code number used for MMCO (Memory Management Control Operation) operation of RPSL (Ref. Another object is to provide a video coding method that enables coding to be performed.

In order to achieve the above object, a video coding method according to the present invention is a field used for RMPNI (Re-Mapping of Picture Numbers Indicator) calculation of a reference picture selection layer (RPSL) in a video coding system using multiple reference pictures. And a code number indicates a picture number information of a target stored in a short-term buffer, a field represented by a DELTA value having a sign, and a code number thereof; A field for indicating picture number information of a target stored in a long-term buffer and a code number thereof; And a field for indicating the end of the RMPNI operation and a code number thereof; Performing video coding using the set result; Its features are to include.

In addition, in order to achieve the above object, the video coding method according to the present invention is a DPN used for MMCO (Memory Management Control Operation) operation of a Reference Picture Selection Layer (RPSL) in a video coding system using multiple reference pictures. Setting a (Difference Picture Numbers) field to a DPN value having a sign; Performing video coding using the set result; Its features are to include.

According to the present invention, the amount of bits can be reduced by saving the codeword length used for the RMPNI operation, and the amount of bits can be reduced by saving the codeword length used for the MMCO operation. There is an advantage that the efficient video coding can be performed by reducing the amount of bits.

Hereinafter, the embodiment according to the present invention will be described in detail.

First, a process of obtaining an ADPN field of a conventional RMPNI operation from an encoder and a decoder of a video coding system is as follows.

[Encoder]

DELTA = PNQ-PNP

If (DELTA <0) {

If (DELTA <-MAX_PN / 2-1)

MDELTA = DELTA + MAX_PN;

else

MDELTA = DELTA;

}

else {

If (DELTA> MAX_PN / 2)

MDELTA = DELTA-MAX_PN;

else

MDELTA = DELTA

}

ADPN = abs (MDELTA);

[Decoder]

If (RMPNI = '1') {

If (PNP-ADPN <0)

PNQ = PNP-ADPN + MAX_PN;

else

PNQ = PNP-ADPN;

}

else {

If (PNP + ADPN> MAX_PN-1)

PNQ = PNP + ADPN-MAX_PN;

else

PNQ = PNP + ADPN;

}

At this time, the codeword of the ADPN is determined by the code number having the value of ADPN-1, as described in the prior art.

In the present invention, in order for the ADPN to have a sign, DELTA will be used instead of the term ADPN. In addition, the present invention, in a video coding system using multiple reference pictures, the field and code number used for the Re-Mapping of Picture Numbers Indicator (RPMNI) operation of the Reference Picture Selection Layer (RPSL) is a short-term buffer (short-term) a field indicating information of a picture number of a target stored in a buffer) and indicated by a DELTA value having a sign and a code number thereof; A field for indicating picture number information of a target stored in a long-term buffer and a code number thereof; And a field for indicating the end of the RMPNI operation and a code number thereof.

One example of the process of obtaining the DELTA in the encoder and the decoder in the video coding system according to the present invention is as follows.

[Encoder]

DELTA = PNQ-PNP;

If (DELTA <-MAX_PN / 2)

DELTA = DELTA + MAX_PN;

If (DELTA> MAX_PN / 2)

DELTA = DELTA-MAX_PN;

[Decoder]

PNQ = PNP + DELTA

If (PNQ <0)

PNQ = PNQ + MAX_PN;

If (PNQ> MAX_PN-1)

PNQ = PNQ-MAX_PN;

Here, PNQ is a target picture number, PNP is a picture number to be predicted, and MAX_PN represents a maximum value of a picture number, respectively.

In addition, the codeword of DELTA obtained by the encoder can be derived from the code table shown in [Table 4].

Code number Number of bits DELTA value 0 One One One 3 -One 2 3 2 3 5 -2 4 5 3 5 5 -3 6 5 4 7 7 -4 8 7 5 ... ... ...

As such, instead of the conventional ADPN field, when specifying a default index using RMPNI, the picture specification of the short-term buffer having the target PN may be represented by a DELTA field having a sign. Accordingly, the codeword of the RMPNI instruction is simplified as shown in the code table shown in [Table 5].

RMPNI code number field 0 Signed DELTA value One LPIR representing LPIN 2 End

Then, the operation bit amount for the case where the value of ADPN is 3, which is performed by the reference picture calculation method in the RPSL according to the prior art, and the DELTA value, which is performed by the reference picture calculation method in the RPSL according to the present invention. Let's compare the operation bit amount for the case of 3.

First, referring to the conventional operation bit amount, when the value of the ADPN is 3, the total operation bit amount is determined by the codeword (3 bits) representing the sign of the ADPN value and the codeword (3 bits) representing the value 3. Will be 6 bits. On the other hand, when looking at the operation bit amount according to the present invention, when the DELTA value is 3, the codeword (0 bits) indicating the sign of the DELTA value is not necessary, and the total is determined by the codeword (5 bits) indicating the value 3. The operation bit amount is 5 bits. As described above, according to the reference picture calculation method in the RPSL according to the present invention, the total amount of computation bits is reduced, so that efficient coding can be performed.

On the other hand, as mentioned in the prior art, since the DPN field used in the conventional MMCO operation is determined by a code number having a codeword equal to that of the DPN, the DPN value should always be positive. In the present invention, when obtaining a DPN, the same procedure applies to obtaining a DELTA.

That is, in a video coding system using multiple reference pictures, a difference picture number (DPN) field used for a Memory Management Control Operation (MMCO) operation of a reference picture selection layer (RPSL) is represented by a DPN value having a sign. It was made. However, since the DPN may have a value of 0, the allocated codeword is modified as shown in [Table 6] and applied to the DPN.

Code number Number of bits DPN value 0 One 0 One 3 One 2 3 -One 3 5 2 4 5 -2 5 5 3 6 5 -3 7 7 4 8 7 -4 ... ... ...

On the other hand, in applying the DPN according to the present invention can be performed through the following coding process.

[Encoder]

DPN = PNC-PNQ;

If (DPN <-MAX_PN / 2)

DPN = DPN + MAX_PN;

If (DPN> MAX_PN / 2)

DPN = DPN-MAX_PN;

[Decoder]

PNQ = PNC-DPN;

If (PNQ <0)

PNQ = PNQ + MAX_PN;

If (PNQ> MAX_PN-1)

PNQ = PNQ-MAX_PN;

Here, PNQ is a target picture number, PNC is a picture number currently being coded, and MAX_PN represents a maximum value of a picture number, respectively.

Even in the case of performing an MMCO operation using a signed DPN value as described above, as in the case of performing an RMPNI operation using the signed DELTA value, it is possible to perform efficient video coding by reducing the total amount of operation bits. It becomes possible.

As described above, according to the video coding method according to the present invention, a DELTA including a code is obtained instead of the conventional ADPN, and a new code table is proposed for the codeword allocation of the DELTA. Also, since DELTA itself contains a sign, the RMPNI operation has been greatly simplified to consist of only three code numbers: the DELTA field, the LPIR field, and the end. In addition, the DPN used for the MMCO operation is different from the conventional case where only positive numbers are used, and a code table for allocating codewords of the DPN is provided by including a sign.

According to the video coding method according to the present invention, it is possible to save the length of the codeword used in the RMPNI operation to reduce the bit amount, and also to save the length of the codeword used in the MMCO operation By reducing the amount of bits, there is an advantage that efficient video coding can be performed by reducing the amount of bits used in the RPSL as a whole.

Claims (7)

In a video coding system that uses multiple reference pictures, In setting the field and code number used for the Re-Mapping of Picture Numbers Indicator (RPMNI) operation of the Reference Picture Selection Layer (RPSL) , it informs the picture number information of the target stored in the short-term buffer. a field represented by a DELTA value having a (sign) and a code number thereof; A field for indicating picture number information of a target stored in a long-term buffer and a code number thereof; And a field for indicating the end of the RMPNI operation and a code number thereof ; Performing video coding using the set result; Video coding method comprising a. The method of claim 1, The field and code number used for the RMPNI operation is set as shown in the following table . RMPNI code number field 0 Signed DELTA value One LPIR representing LPIN 2 End The method of claim 1, The DELTA value, code number, and number of bits used for the RMPNI operation are set as shown in the following table . Code number Number of bits DELTA value 0 One One One 3 -One 2 3 2 3 5 -2 4 5 3 5 5 -3 6 5 4 7 7 -4 8 7 5 ... ... ... The method of claim 1, The DELTA value used for the RMPNI operation is processed in an encoder and a decoder of a video coding system as follows . [Encoder] DELTA = PNQ-PNP; If (DELTA <-MAX_PN / 2) DELTA = DELTA + MAX_PN; If (DELTA> MAX_PN / 2) DELTA = DELTA-MAX_PN; [Decoder] PNQ = PNP + DELTA If (PNQ <0) PNQ = PNQ + MAX_PN; If (PNQ> MAX_PN-1) PNQ = PNQ-MAX_PN; Here, PNQ is a target picture number, PNP is a picture number to be predicted, and MAX_PN represents a maximum value of a picture number, respectively. In a video coding system that uses multiple reference pictures, Setting a DPN (Difference Picture Numbers) field used for Memory Management Control Operation (MMCO) operation of a Reference Picture Selection Layer (RPSL) to a DPN value having a sign ; Performing video coding using the set result; Video coding method comprising a. The method of claim 5, The DPN value, code number, and number of bits used for the MMCO operation are set as shown in the following table . Code number Number of bits DPN value 0 One 0 One 3 One 2 3 -One 3 5 2 4 5 -2 5 5 3 6 5 -3 7 7 4 8 7 -4 ... ... ... The method of claim 5, The DPN value used for the MMCO operation is processed in the encoder and the decoder of the video coding system as follows . [Encoder] DPN = PNC-PNQ; If (DPN <-MAX_PN / 2) DPN = DPN + MAX_PN; If (DPN> MAX_PN / 2) DPN = DPN-MAX_PN; [Decoder] PNQ = PNC-DPN; If (PNQ <0) PNQ = PNQ + MAX_PN; If (PNQ> MAX_PN-1) PNQ = PNQ-MAX_PN; Here, PNQ is a target picture number, PNC is a picture number currently being coded, and MAX_PN represents a maximum value of a picture number, respectively.