JPH0767105A - Image encoding method - Google Patents

Abstract

PURPOSE:To reduce the amount of data to be transmitted by preventing useless data from being transmitted as much as possible by forcedly defining an isolate significant block, for which all the surrounding blocks are non-significant blocks, as the non-significant block in the case of such a significant block. CONSTITUTION:The picture of one frame in image information is horizontally and vertically divided into plural blocks, difference between frames is calculated for each block, that difference is compared with a threshold value for each block, it is decided whether the respective blocks are non-significant blocks or significant blocks, and only the image data of the significant blocks are encoded and transmitted. In this case, at the time of the isolate significant block for which the target block of a coordinate (4, 3) is the significant block and all the 24 surrounding blocks are the non-significant blocks, that target block is forcedly changed into the non-significant block. By providing such a surrounding block reference/decision circuit, the amount of data to be transmitted is reduced.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding method suitable for application to, for example, hybrid coding in which interframe coding and orthogonal transform coding are combined.

[0002]

2. Description of the Related Art A system for conducting a video conference using an ISDN line is known. In this video conference system, the image data is compressed and encoded and transmitted so that the image information can be transmitted through the ISDN line. As this image coding method, CCITT standard H.264 is used. The hybrid coding method adopted in H.261 is used.
This hybrid coding system is basically a combination technique of interframe coding and orthogonal transform coding (DCT coding in this case).

In this hybrid coding system,
As shown in FIG. 3, horizontal direction × vertical direction = 352 pixels ×
The image information of one frame consisting of 288 lines is divided into blocks BLK of DCT processing unit consisting of 8 pixels × 8 lines, and the current frame F is divided into blocks BLK.
The difference between the information of L and the information of the frame DFL one frame before is obtained at the corresponding positions. Then, the difference value ΔBLK for each block BLK is DCT encoded and transmitted to the receiving side.

By the way, in this hybrid coding system, not all image data is transmitted to the receiving side, but image data having no difference or a slight block between preceding and following frames is not transmitted to the receiving side. . This is because, on the receiving side, when the relevant block portion of the current frame to be reproduced has no difference or is a slight block, the image information of the previous frame can be used as it is instead of the image information.

Therefore, the difference value for each block BLK is
When the difference value is larger than the threshold value Th, which is one of the parameters of the encoding control, and the difference value is larger than the threshold value Th, it is determined that the block of the current frame is a significant block, and when the difference value is smaller than the threshold value Th, The block performs a determination operation to determine that it is an insignificant block. Then, if the determination result is a significant block, the procedure for transmitting the block to the receiving side is taken, but when the block is an insignificant block,
No data is transmitted to the receiving side.

Further, in this hybrid coding system,
Encoding is performed in units of blocks BLK, but the transmission units when actually transmitting image information from the transmission side to the reception side are different, and the minimum unit of transmission is a macroblock that is a collection of a plurality of blocks. The transmission data has a hierarchical structure.

FIG. 4 shows a hierarchical structure when transmitting block information. That is, as shown in FIG. 4D,
One block B consisting of 8 pixels x 8 lines = 64 pixels
The LK is subjected to orthogonal transformation (DCT) of the difference value ΔBLK to be transformed into block data BD including 64 pieces of coefficient data AC sequentially arranged from the one having a lower spatial frequency to the one having a higher spatial frequency.

As described above, this block data BD
Is a unit of DCT encoding, but transmission of image information from the transmitting side to the receiving side is physically performed by macroblock MB which is a total of four blocks BD, two in the horizontal direction and two in the vertical direction. (See FIG. 4C) is the minimum unit. Note that, as shown in FIG. 4C, the macroblock MB is substantially composed of a total of 6 blocks, that is, four blocks of the luminance signal component Y and two blocks of the color difference signal components (CR, CB). ing.

11 macro blocks MB in the horizontal direction,
A group of blocks GOB (see FIG. 4B) is a group of 33 pieces in total in the vertical direction. And
The group of blocks GOB, two in the horizontal direction and six in the vertical direction, are arranged in one screen data CIF.
(See FIG. 4A).

FIG. 5 shows a format for transmitting data of each layer. The meaning of each data in FIG.
It is as follows, and the numerical value under each data shows the bit length of the data. The data indicated by "*" as the bit length indicates that it is a variable length bit.

PSC frame start number TR frame number PTYPE type information PEI extension data insertion information PSPARE preliminary information GBSC GOB start code GN GOB number GQUANT quantization characteristic information GEI extension data insertion information GSPARE preliminary information MBA macroblock address MTYPE macro block Type information MQUANT Quantization characteristic information MVD Motion vector information CBP Significant block pattern information TCOEFF Transform coefficient EOB block end code The type information of "MTYPE" is the type of arrangement pattern of significant blocks in macroblocks and insignificant blocks. Is indicated by a pattern number.

In the format of the transmission data shown in FIG. 5, when the macroblock MB has one or more significant blocks, the data format of the macroblock MB is the third-stage format MB (A) from the top. On the other hand, a macroblock MB consisting of all insignificant blocks
The transmission data format is the lowest format MB (B).

Comparing the two formats, in the case of the format MB (B), as compared with the format MB (A), the data [TCOEFF], [EO] of the block BLK.
It can be seen that B], [MQUANT], and [CBP] are not transmitted, and the amount of transmission data is considerably reduced. That is, in the hybrid coding method, the image data of the insignificant block is not transmitted, and the macro block MB
The transmission format of 1 is changed depending on whether it includes one or more significant blocks or when all of them consist of insignificant blocks, so that the transmission data amount is further reduced.

[0014]

By the way, as described above, in the conventional hybrid coding system, each block BLK is individually treated to make a significant or insignificant decision. For this reason, when most of the surroundings are insignificant blocks, the image data of the significant block is transmitted even if the significant block exists in isolation. However, such isolated significant blocks are often noise. In addition, even if it is not noise, when encoding at a high compression rate such as 1/1000, the isolated significant block often does not have a significant meaning.

Further, as described above, in the case of the conventional hybrid coding system, since the data is transmitted in macroblock units, only when all of the four blocks BLK included therein are insignificant blocks. Data is transmitted in the format MB (B) having a small data amount.

Therefore, only one block BLK among the four blocks constituting the macroblock is significant,
In addition, even in the case where the significant blocks are isolated such that the surroundings of the block are all insignificant blocks, the macro block MB including the block BLK determined to be the significant block has a large amount of data. It had to be transmitted in format MB (A).

In view of the above points, an object of the present invention is to provide an image coding method capable of reducing the amount of transmitted data by transmitting meaningless data as little as possible. To do.

[0018]

In order to solve the above-mentioned problems, an image coding method according to the present invention uses image information as 1
The screen of the frame is divided horizontally and vertically, divided into a plurality of blocks, the difference between the frames is obtained for each block, and the difference is compared with the threshold value for each block, In the image coding method, in which each block is a significant block or an insignificant block is determined, and only the image data of the significant block is encoded and transmitted, refer to blocks around the block determined to be a significant block, When all of these surrounding blocks are insignificant blocks, the block determined to be the significant block is forcibly changed to an insignificant block.

[0019]

In the present invention having the above-described structure, even if a significant block is referred to, its surrounding blocks are referred to, and when the surrounding blocks are all isolated significant blocks that are insignificant blocks, they are forcibly forced. The significant block is regarded as an insignificant block. Therefore, the image data of the isolated significant block is not a transmission target, and the amount of transmission information can be reduced.

[0020]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the image coding method according to the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram of an embodiment of an image coding apparatus which realizes the method according to the present invention. This example is based on the CCITT standard H.264. 26
This is an example in the case of the hybrid coding method adopted in No. 1.

The digital image signal is supplied to the subtraction circuit 12 through the input terminal 11. The digital image signal from the input terminal 11 is also supplied to the image memory 18,
It is delayed by one frame period. The image memory 18 has a variable delay function for motion compensation, as will be described later.

The digital image signal of one frame before from the image memory 18 is supplied to the subtraction circuit 12 via the in-loop filter 19. The in-loop filter 19 can be turned on / off, and the on / off signal f is supplied to the video signal multiplexing unit 21. The subtraction circuit 12 obtains the difference value between frames for each block BLK as described above.

The block B which is the output of the subtraction circuit 12
The inter-frame difference value for each LK is supplied to the converter 13 via the switch circuit SW1 and DCT conversion is performed. The output of the converter 13 is supplied to the quantizer 14, from which the quantized output index q of the transform coefficient is obtained and supplied to the video signal multiplexer 21. The quantization characteristic designating signal qz is supplied from the coding controller 20 to the quantizer 14, and the characteristics at the time of quantization are designated.

The quantized output index q from the quantizer 14 is also supplied to the inverse quantizer 15 and inversely quantized, and then supplied to the inverse DCT converter 16 to obtain the block BLK.
The difference value for each is reproduced. Then, the reproduced difference value is supplied to the image memory 18 through the addition circuit 17. When the in-loop filter 19 is turned on, the switch circuit SW2 is switched and controlled by the switching signal from the encoding controller 20 to the state shown in the figure. The image data of one frame before is supplied.

In the image memory 18, the motion vector v is detected from the signal of one frame from the adder circuit 17 and the signal of the current frame from the input terminal 11, and this is supplied to the video signal multiplexer 21. Using this, the input image signal is motion-compensated to obtain an image signal delayed by one frame.

The switch circuit SW1 is for switching between the case where the interframe difference is obtained as the image data and the DCT encoding is performed, and the case where the image data is directly DCT encoded without obtaining the difference. The switching signal is supplied from the encoding controller 20.
Then, an identification flag p indicating which mode has been adopted is provided from the encoding controller 20 to the video signal multiplexer unit 21.
Supply to.

The above-mentioned structure is the same as that of the conventional image coding apparatus of the hybrid coding system. In the present invention, in addition to the above configuration, the peripheral block reference / determination circuit 3
0 is set.

The difference value for each block BLK from the subtraction circuit 12 is supplied to the determination circuit 30. Circuit 30
In, the difference value for each input block BLK is
The block BLK is supplied to the significance / insignificance determination circuit 31.
For each time, the difference value and the threshold value Th are compared, and a flag S indicating whether the difference value of each block BLK is significant or insignificant
F is required. The significant / insignificant flag SF for each difference value of each block BLK is supplied to and stored in the shift register 32.

The flags SF of a plurality of peripheral blocks centered on the target block, which are stored in the shift register 32, are supplied to the peripheral reference determining circuit 33. The surrounding reference determination circuit 33 has a width of two blocks around the target block BLK, in this example, two blocks vertically and two blocks horizontally.
Even if the target block is a significant block, it is determined whether or not the target block is a significant block by referring to the flags SF for a total of 24 blocks of the block width. The reference determination method will be described with reference to FIG.

That is, in the present invention, each block BL
Referring to blocks around K, in this example, as shown in FIG. 2, the macroblock MB including the target block
The blocks within the eight macroblocks around are referred to. Then, within the macroblock, 5 × 5 blocks centered on the target block are used for reference. Therefore, the shift register 32 is
2/8 × 5 lines = 220 bits capacity is enough. In FIG. 2, the crosses indicate blocks that are not referred to within the eight macroblocks around the macroblock MB including the target block.

The determination as to whether the target block is an isolated significant block is included in the eight macroblocks around the macroblock including the target block when the target block is a significant block in this example. It is determined from the flags SF whether or not all of the 24 blocks around the target block are insignificant blocks.

In FIG. 2, an insignificant block is shown when the flag SF is "0", and a significant block is shown when the flag SF is "1". In the example of FIG. 2, when the coordinates (x, y) of the upper left block are (1, 1), the coordinates (2,
1) to (6,1), (2,2) to (6,2), (2
3), (3,3), (5,3), (6,3), (2
4) to (6,4) and (2,5) to (6,5) are insignificant blocks. Further, the block of coordinates (4, 3) is the target block and indicates that it is a significant block.

That is, in the state of FIG. 2, coordinates (4, 3)
The target block of is a significant block, the 24 blocks around it are all insignificant blocks, and the target block is in a state of being an isolated significant block.

The judgment output of the judgment circuit 33 is supplied to the flag forced change circuit 34. In addition, the shift register 3
From 2, the flag SF of the target block is supplied to the flag compulsory changing circuit 34. In the flag compulsory change circuit 34, when the judgment result of the surrounding reference judgment circuit 33 indicates that the target block is an isolated significant block, the flag is compulsorily changed from "1" to "0", Forcibly change to the flag of the unwilling block.

In other cases, that is, when the target block is not an isolated significant block, the flag forced change circuit 3
From 4, the flag of the target block is output as it is. The flag output from the flag compulsory change circuit 34 is
It is supplied to the transmission / non-transmission flag setting circuit 35.

The transmission / non-transmission flag setting circuit 35 determines whether the macro block is transmitted or not, and sets the flag. For a macroblock including an isolated block, the flag of the isolated significant block has been changed to a flag indicating that it is an insignificant block, so it is determined that the macroblock is not transmitted, and the flag is
Not transmitted. This transmission / non-transmission flag is supplied to the video signal multiplexing unit 21 as the flag t via the encoding controller 20.

The video signal multiplexing unit 21 forms a signal in the above-mentioned transmission format based on the flags and other signals supplied thereto. In this case, the macroblock including the isolated significant block as shown in FIG.
The data is non-transmission data and is transmitted not in the format MB (A) but in the format MB (B) having a small amount of transmission data.

As described above, when there is only one significant block in the macroblock MB and the significant block is an isolated significant block, the macroblock is forcibly made non-transmitted data. As a result, transmission data can be reduced. By using the reduced information for other encoding, the overall encoding efficiency can be improved. Even if the transmission data is reduced in this way, isolated significant blocks are often noise, and even if they are not noise, the isolated significant blocks are transmitted in the image coding method with a large compression rate. Even without it, the effect on image quality is small.

The above example is based on CCITT standard H.264. Two
This is a case where the present invention is applied to the 61-compliant hybrid coding system, but it goes without saying that the present invention is not limited to this and can be applied to other coding systems.

As a method for determining whether or not the target block is an isolated significant block, as shown in FIG. 2, reference is made to surrounding blocks in the range of two blocks in each of the vertical direction and the horizontal direction. However, as the range of these reference blocks, it is possible to set an arbitrary range in consideration of conditions such as image quality.

In the example of FIG. 1, the method of the present invention is
Although it is realized by hardware, it can be realized by software using a microcomputer.

[0042]

As described above, according to the present invention, in encoding, in a method of judging whether each block is significant or insignificant and determining whether or not to transmit, the surroundings of the target block are determined. When the target block is an isolated significant block by referring to the block, the target block is regarded as an insignificant block and is not transmitted, so that it is possible to reduce the amount of transmission data. There is also an advantage that the coding efficiency is improved as a whole by utilizing the reduced information capacity for other coding.

Since an isolated significant block is often generated by noise, it is possible to expect a noise removal effect by forcibly changing it to an insignificant block.

When applied to the hybrid coding method, a macroblock is a transmission unit, and the transmission format is different when the macroblock is made up of all insignificant blocks and when there is at least one significant block. Since the transmission format of macroblocks, which consist of all insignificant blocks, has a significantly small amount of transmission data,
The effect of this invention is greater.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of an image encoding apparatus that realizes an image encoding method according to the present invention.

FIG. 2 is a diagram for explaining an embodiment of an image coding method according to the present invention.

FIG. 3 is a diagram for explaining a hybrid coding method.

FIG. 4 is a diagram for explaining a hierarchical structure of transmission data in the hybrid coding method.

[Fig. 5] Fig. 5 is a diagram for explaining a format of transmission data in a hybrid encoding method.

[Explanation of symbols]

 12 subtraction circuit 13 DCT converter 14 quantizer 18 image memory for delay 20 encoding controller 21 video signal multiplexer 30 peripheral block reference / judgment circuit 31 significant / intentional judgment circuit 32 shift register 33 surrounding reference judgment circuit 34 Flag forced change circuit 35 Transmission / non-transmission flag setting circuit

Continuation of front page (51) Int.Cl. ⁶ Identification number Office reference number FI technical display location H04N 11/04 A 7337-5C

Claims (3)

[Claims] 1. Image information is divided into a plurality of blocks by dividing a screen of one frame in a horizontal direction and a vertical direction, a difference between frames is obtained for each block, and the difference is calculated for each block. In comparison with the threshold value, it is determined whether each block is a significant block or an insignificant block, and in the image coding method in which only the image data of the significant block is encoded and transmitted, it is determined as a significant block. Block around the specified block, and when all these blocks are unwilling blocks,
An image coding method characterized in that the block determined to be the significant block is forcibly changed to an insignificant block. 2. The image coding method according to claim 1, wherein the image information is transmitted for each transmission unit block including a plurality of the blocks, which is a unit of encoding, and the transmission unit block When there is only one significant block in the block, reference is made to the surrounding blocks of the significant block, and when all of these surrounding blocks are insignificant blocks, the significant block is forcibly changed to an insignificant block. An image encoding method characterized in that a transmission unit block including a significant block of blocks is transmitted in the same transmission format as when all blocks are insignificant blocks. 3. The image coding method according to claim 2, wherein the surrounding blocks to be referred to include blocks in transmission unit blocks around a transmission unit block including the significant block of the one block. Characteristic image coding method.