JP2004112842A - Coding device for coding block pattern and coding method for coding block pattern - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a data compression method which enhances the compression rate of transmitted data. <P>SOLUTION: When an object is selectively present in unit blocks in a macro block, variable length code tables for block patterns mutually different depending on the number of the unit blocks in which the object is in existence is applied. <P>COPYRIGHT: (C)2004,JPO

Description

The present invention relates to an encoding device and an encoding method for converting video data in a frame into variable-length coded data and transmitting the data.

In recent years, videophones, video conferences, and the like have been provided using an ISDN (Integrated Service Digital Network). H.261 for videophone services using the Public Switched Telephone Network (PSTN). H.263 and the like have been proposed as technical standards for global standardization of image compression technology.

H. 提供, which offers compression technology for ultra-low-speed video, such as a videophone 261 and H.E. In H.263, it is known that the quality of an image is degraded, and an improvement for upgrading the image quality is required. On the other hand, moving picture related compression techniques include MPEG1 (Motion Picture Expert Group) for DSM (Digital Storage Media), DSM, HDTV (High Density Television), MPEG2 for ATV (Advanced Television: next generation television) and the like has been standardized.

These H. 261, H .; 263, MPEG1 and MPEG2 compression techniques employ block-based encoding methods. However, when these encoding methods are applied to ultra-low-speed transmission, image quality is noticeably reduced. For this reason, a completely new method has been proposed as MPEG4, which is different from the existing method, and the standardization of block-based coding in which the concept of an object is introduced is progressing.

Of these, the object-oriented coding (Object Oriented Coding) is based on an object (Object) in a region that moves between two images without using a block as a base, so that the entire image can be divided into a background (Background) and a non- In this method, coding is performed by dividing into a covered region (Uncovered Region), a region where motion compensation is possible (MC-Region), and a region where motion compensation is not possible (MF-Region).
Here, the motion compensable object (MC-object) refers to an object that converts an object in a three-dimensional space into a two-dimensional object and moves according to a certain rule such as horizontal movement, rotational movement, and linear deformation. An impossible object (MF-object) refers to an object to which this certain rule is not applied.

FIG. 7 is a drawing showing the relationship between one frame grid 31 and one unit macroblock 33. In FIG. One frame grid 31 is divided into an array in the X-axis and Y-axis directions, and each division becomes a unit macroblock 33. FIG. 7A shows a frame grid in which the grid start position points are X-0 and Y-0, and FIG. 7B shows one unit macroblock in an enlarged manner. A state where the image is divided into eight on the X axis and eight on the Y axis is shown. Here, the divided element is referred to as a pixel 34. The unit macro block 33 is called an N × M macro block according to the number of pixels. Therefore, the unit macroblock in FIG. 7B is an 8 × 8 macroblock.

A conventional shape adaptive technique used as a technique for encoding the object thus divided will be considered with reference to FIG.
Here, the frame grid has 8 × 8 macroblocks, and when all of the objects for which movement cannot be compensated for are filled in the macroblocks, the same as a two-dimensional DCT (Discrete Cosine Transform) is used. Become efficient. If all the objects that cannot be compensated for movement are not satisfied in the block, one-dimensional DCT is performed on the X-axis only for the shape portion of the corresponding object that cannot be compensated for motion, and one-dimensional DCT is performed on the Y-axis.

Hereinafter, such a conventional shape adaptive technique will be described in more detail.
FIG. 8A shows a state in which an object that cannot be compensated for movement (a shaded portion) exists in an 8 × 8 macro block.
In order to encode this non-motion compensable object by the shape adaptive DCT, first, as shown in FIG. Is encoded by one-dimensional DCT.

黒 い The black circles in FIG. 8C represent the discrete cosine transform (DC) values of the one-dimensional DCT in the vertical direction. In this state, one-dimensional DCT in the Y direction is performed. When the one-dimensional DCT in the Y direction is performed as described above and the state shown in FIG. 8D is obtained, the block is again filled with the left boundary of the block as shown in FIG. Perform DCT coding. When the X-direction one-dimensional DCT is completed, the X-direction and Y-direction one-dimensional shape adaptive DCT is completed as shown in FIG.

A zigzag scan (ZIGZAG SCAN) is performed according to FIG. 8F of the final form, and a moving object is scanned diagonally from the uppermost block on the leftmost side.
MPEG-1, MPEG-2, JPEG, and H.264 using the block center coding described above. 261, H .; Existing standards, such as H.263, use macroblocks MB of the form shown in FIG. In this case, four blocks Y1, Y2, Y3, and Y4 exist in one macroblock as shown in FIG. When encoding four blocks (blocks) existing in the macro block MB, there are cases where data exists in each of the blocks Y1, Y2, Y3, and Y4 and cases where data does not exist. Existing standards use a Coded Block Pattern to indicate whether data exists in a block.

FIG. 3 shows a variable length code table for a code block pattern used in the H.263 standard. The variable length code table includes an index column indicating 16 states, a CBPY (intra mode) column indicating a code block pattern (CBPY) in the intra (INTRA) mode, and a code block pattern (CBPY) in the inter (INTER) mode. ), A bit number field indicating the bit length of the variable length coded data, and a variable length coded data field indicating the variable length coded data.

Also, (1, 2, 3, 4) in the CBPY column correspond to four unit blocks Y1, Y2, Y3, and Y4 in which data exists, where 1 is Y1, 2 is Y2, 3 is Y3, 4 corresponds to Y4. Therefore, for example, the index 1 indicates that the value of CBPY (intra mode) is represented by (0, 0, 0, 1) because there is no data in blocks Y1, Y2, and Y3, and in block Y4. Indicates that data exists. In this case, the variable-length coded data is 5 bits and is represented by 00101. By using the variable length code table in this way, it is possible to effectively indicate whether or not data exists in each macroblock, and to transmit using the variable length coded data obtained by the conversion. The amount of data can be reduced, that is, data compression can be realized.

However, in such a conventional data compression method using block-centered video signal encoding, a code block pattern is used to indicate whether or not data exists in a macroblock. Since the length code table is fixed to one, there is a problem that the reduction efficiency of the transmitted data is reduced.

In other words, the number of objects is different from each other and distributed in one macroblock (for example, even if a small object exists only in a part of a macroblock or an object exists in the entire macroblock), it is the same. , The data reduction efficiency is reduced.
An object of the present invention is to solve the problem of a decrease in coding efficiency that occurs during the conventional object-centered coding, and an object of the present invention is to provide a data compression method that increases the reduction efficiency of transmitted data.

The coding block pattern coding apparatus of the present invention uses the coded shape information to generate four variable-length codes according to the number of unit blocks in which an object exists in a macroblock having a plurality of unit blocks. Control means for selecting one of the tables and selecting a variable length code on the variable length code table selected according to the coding block pattern of the macroblock to be coded; A variable-length code table storing means for storing the four variable-length code tables corresponding to the number of unit blocks in which the object exists, and the variable-length code code according to a coded block pattern in the macro block. And a video signal encoding unit that skips encoding of texture information in a unit block where no object exists. And wherein the Mukoto.

As described above, the present invention applies different variable-length code tables having the highest coding gains according to the presence or absence of an object in a unit block obtained by dividing each macroblock. There is an advantage that the data reduction rate can be increased and the coding gain can be increased.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
A system to which the present invention is applied is a general video encoding system shown in FIG. 5, in which a video signal is input, and a video object plane (VOP) that divides a video of one frame into video by object. Surface) forming unit 10, an encoding unit 20 for encoding shape information of each object obtained from the VOP forming unit 10, and respective encoders 21, 22, 23, ... in the encoding unit 20. And a multiplexing unit 30 for multiplexing the obtained encoded video signals. The VOP forming unit 10 divides the video information of one frame input from the video signal input unit into objects, and the encoders 21, 22, 23,. Encode.

Here, one encoder 21 in the encoding unit 20 is configured as shown in FIG. A switch 21a for selecting object shape information, a shape information encoding unit 21c for encoding the shape information selected by the switch 21a, a shape information encoded by the shape information encoding unit 21c, and arbitrary shape information And a motion estimator 21e for estimating the motion of the shape information selected by the switch 21a using the reproduced previous shape information or information of an arbitrary shape, and a motion estimator 21e. A motion compensator 21f for compensating the motion of the shape information according to the amount of motion to be performed, a subtracter 21g for subtracting the video compensated by the motion compensator 21f from the video obtained by the VOP forming unit 10, and a subtractor 21g. The video signal encoding unit 21h that encodes the obtained video signal, and the video information obtained from the video signal encoding unit 21h and the motion compensated by the motion compensation unit 21f. An adder 21i for adding the compensated video signal, a VOP reproducing unit 21j for reproducing a VOP with the video signal obtained from the adder 21i, and different blocks according to the number of blocks in the macroblock where an object exists. And a memory 21b storing a variable length code table for the pattern.

The encoder 21 configured as described above selects the object shape information obtained from the VOP forming unit 10 by the switch 21a, and the selected shape information is encoded by the shape information encoding unit 21c as the shape information of the corresponding object. Be converted to Then, the coded shape information and the VOP of arbitrary shape information are selectively transmitted to the multiplexing unit 21k by the selection switch 21d, and the motion estimating unit 21e calculates the motion from the shape information obtained by the VOP forming unit 10. The motion compensating unit 21f estimates the motion and compensates for the motion according to the estimated motion amount of the object.

Then, after the motion compensated video is subtracted by the subtractor 21g from the object shape information obtained from the VOP forming unit 10, the video signal is encoded by the video signal encoding 21h to the multiplexing unit 21k as signal information. Is transmitted.
At this time, the video signal encoding unit 21h selectively applies the variable length code table for the different block patterns stored in the memory 21b according to the number of blocks in which the objects in the macroblock are present, to code blocks. Generate a pattern.

Note that the video signal encoded by the video signal encoding unit 21h is added to the motion-compensated video signal by the adder 21i, and is then reproduced by the VOP reproduction unit 21j as a VOP before. The motion estimating unit 21e estimates the motion of the object shape information obtained from the VOP forming unit 10, and transmits the motion information estimated based on the result to the multiplexing unit 21k.

Thereafter, the multiplexing unit 21k multiplexes and outputs the input coded shape information, motion information, and signal information, and outputs the multiplexed signal as a bit stream to a decoder (not shown) via the buffer unit 21m. Transmitted.
Here, the core of the present invention is to selectively apply a plurality of variable length code tables stored in the memory 21b according to the number of objects present in each unit block in the macroblock.

Next, an embodiment of the present invention for this purpose will be described.
That is, as shown in FIG. 1, since the object P does not always form a square, when the object P is divided into macroblock MBs, the blocks including the object P and the blocks not including the object P are included in the divided macroblock MB. Can be.

As shown in FIG. 1A, an object P represented as video information data in one frame divided into a number of macroblocks MB includes a portion existing as data in each divided macroblock and a portion not present. There is.
FIG. 1B is an enlarged view of an area Q surrounded by a dotted line in FIG. The shaded portion in the drawing indicates the object P. FIG. 1B shows four macro blocks MB1, MB2, MB3, and MB4, and each macro block is composed of four unit blocks Y1, Y2, Y3, and Y4.
Therefore, it can be seen that there are unit blocks including the object P and macroblocks that do not include the object P. This is shown in FIG.

That is, the macroblock 1 (MB1) shown in FIG. 2A has all objects in the other unit blocks Y2, Y3, and Y4 except for the first unit block Y1, and the macroblock B shown in FIG. In block 2 (MB2), all objects exist in all unit blocks Y1, Y2, Y3, and Y4. In macroblock 3 (MB3) in (C), objects exist only in the second unit block Y2. In the macroblock 4 (MB4) of (D), an object exists only in the first unit block Y1 and the second unit block Y2.

(4) Since an object is selectively present in each unit block in the macroblock, different variable length code tables are applied to each case to increase the coding gain and increase the data reduction efficiency.

That is, since an object exists only in three unit blocks Y2, Y3, and Y4 in macro block 1 (MB1), if a variable length code table as shown in Table 1 is applied, data reduction becomes possible and encoding is performed. Gain can be increased.

In macro block 2 (MB2), objects exist in all four unit blocks Y1, Y2, Y3, and Y4. In this case, the existing variable length code table shown in FIG. Since an object exists only in one unit block Y2 in MB3), a variable length code table as shown in Table 3 is applied, and an object exists only in two unit blocks Y1 and Y2 in macro block 4 (MB4). Therefore, a variable length code table as shown in Table 2 is applied.

Note that, in the above-described embodiment, a case has been described in which each macroblock is divided into four square unit blocks, and different variable-length code tables are applied. However, the number of divisions is not limited to this. When the number of divisions is adopted, different variable-length code tables having the highest encoding gain can be similarly obtained by experiments according to the presence or absence of an object in each unit block.
By applying different variable-length code tables (variable-length code tables having the best coding gain) according to the presence or absence of an object in each unit block in a macroblock, the data reduction rate and the coding gain can be reduced. Can be enhanced.

Here, the different variable-length code tables shown in Tables 1, 2, and 3 are new variable-length code tables obtained by experiments. Here, the terms of intra (INTRA) and inter (INTER) are as follows. Is defined.
The characteristics of the object on the time axis are the case of an object that was not present in the previous frame but appeared in the current frame (object type I) and the case of an extremely small object (object within 10 pixels; object type II) And the case of a relatively extremely large object (object of 10 pixels or more; object type III). Here, the case of an object which did not exist in the previous frame but appeared in the current frame is referred to. In the case of an extremely small object having a size of less than 10 pixels, the coding type is determined to be INTRA mode (non-prediction mode), and in the case of an object not having the above characteristics, the coding type is determined to be INTER mode (prediction mode). .

FIG. 3 is an exemplary diagram in which an arbitrary object is divided in units of macro blocks. The figure which displayed the unit block in which the object exists in a macroblock. MPEG-1, MPEG-2, H.264; 261, H .; FIG. 6 is a diagram showing a macroblock MB including four unit blocks Y1, Y2, Y3, and Y4 used in an existing standard such as H.263. H. 263 is a variable-length coding table for a code block pattern used in the standard. 1 is a block diagram of an object video coding system to which the present invention is applied. FIG. 6 is a detailed configuration diagram of one encoder 21 in FIG. 5. FIG. 3 is an explanatory diagram of a general frame grid and macro blocks. FIG. 3 is an explanatory diagram of a general shape adaptive DCT transform.

Explanation of reference numerals

Reference Signs List 10 VOP forming unit 20 Encoding unit 30 Multiplexing unit MB1 to MB4 Macroblock Y1 to Y4 Unit block

Claims (11)

Utilizing the encoded shape information, one of the four variable length code tables is selected according to the number of unit blocks in which an object is present in a macroblock having a plurality of unit blocks, and is encoded. Control means for selecting a variable length code on the variable length code table selected according to the coding block pattern of the macro block,
Variable length code table storage means for storing the four variable length code tables corresponding to the number of unit blocks in which objects are present in the macroblock;
Video signal encoding means for transmitting the variable length code according to the encoded block pattern in the macroblock and omitting encoding of texture information in a unit block where no object exists. An encoding device for an encoded block pattern. The variable length code table storage means stores a first variable length code table corresponding to a macro block including a unit block in which four objects exist,
The first variable length code table includes:

2. The apparatus according to claim 1, wherein: The variable length code table storage means stores a second variable length code table corresponding to a macroblock including a unit block in which three objects exist;
The second variable length code table includes:

2. The apparatus according to claim 1, wherein: The variable length code table storage means stores a third variable length code table corresponding to a macroblock including a unit block where two objects exist,
The third variable length code table is:

2. The apparatus according to claim 1, wherein: The variable length code table storage means stores a fourth variable length code table corresponding to a macro block including a unit block where one object exists,
The fourth variable length code table is:

2. The apparatus according to claim 1, wherein: A first step of determining the number of unit blocks where an object is present in a macroblock having a plurality of unit blocks and an encoded block pattern of the encoded macroblock using the encoded shape information;
A second step of selecting one of the four variable-length code tables according to the number of unit blocks in which the object exists;
A third step of selecting a variable-length code on a variable-length code table selected according to a coding block pattern of the macroblock to be coded;
A fourth step of transmitting the variable length code and omitting the encoding of texture information for a unit block in which no object is present in a macroblock. The first step is
Detecting a unit block in which at least one object is present in a macroblock using the encoded shape information;
Counting the number of detected unit blocks;
7. The method of claim 6, further comprising: determining the coded block pattern based on the counted number of unit blocks. If the macroblock includes a unit block in which four objects exist, a first variable length code table is selected,
The first variable length code table includes:

The encoding method of an encoded block pattern according to claim 6, wherein A second variable length code table is selected when the macroblock includes a block in which three objects exist;
The second variable length code table includes:

The encoding method of an encoded block pattern according to claim 6, wherein A third variable length code table is selected when the macroblock includes a block in which two objects exist;
The third variable length code table is:

The encoding method of an encoded block pattern according to claim 6, wherein A fourth variable-length code table is selected when the macroblock includes a block in which one object exists;
The fourth variable length code table is:

The encoding method of an encoded block pattern according to claim 6, wherein

Images