JPS6342588A - Image signal encoding method - Google Patents

Abstract

(57) [Abstract] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Summary] When an image signal is encoded and transmitted, it is encoded by adding an orthogonal transform operation, and a specific isolated significant coefficient among the significant coefficients after the orthogonal transform is By excluding the data from the encoding target, it is possible to compress the bandwidth, that is, improve the transmission efficiency.

[Industrial application field]

The present invention relates to a band compression encoding method for transmitting an image signal at a low bit rate, and more specifically, to a band compression encoding method that employs a method of encoding an image signal by adding an orthogonal transform operation.

Among the functional parts in the image signal processing device, the encoding unit occupies a fairly important position. This is because the quality of the image on the receiving side is greatly influenced by the quality of this encoding section. However, there is a contradictory relationship between improving image quality and improving transmission efficiency (the former requires a high bit rate, and the latter requires a low bit rate), and encoding using so-called band compression can meet the contradictory requirements of the two. It has been widely adopted as satisfactory. Among these, the present invention is based on a method of performing band compression by encoding using orthogonal transform.

[Conventional technology]

FIG. 7 is a block diagram showing an example of a general image signal processing device. In this figure, the image signal Sv applied to the input terminal 11 is subjected to a two-dimensional orthogonal transformation operation in a two-dimensional orthogonal transformer 12, and the data subjected to the two-dimensional orthogonal transformation is converted into a two-dimensional significant coefficient in a quantizer 13. and O coefficients, and after the entropy coder 14 assigns a code vl to each quantization level (significant coefficient, 0 coefficient), the output end 15 outputs the band-compressed coded signal Sc. Sent to the transmission path.

The orthogonal transformer 12 described above performs so-called cosine transform, and obtains two-dimensional orthogonal transform data in the form of taking the frequency spectrum of the image signal 3v. As a result, data of 64 pixels, for example, is reduced to data of about 20 significant coefficients and 0 coefficients.

FIG. 8 is a diagram showing a known form of cosine transformation. In this figure, 21 is block data of the image signal SV as it is, and 22 is block data after cosine transformation of the image signal Sv. For example, the image signal SV is 30 times per second.
This method forms each frame that changes frame by frame, but in this orthogonal transformation method, one frame is usually divided into, for example, 32x32
Divide into (vertical x horizontal) blocks and process each block. Each block is composed of 8×8 (vertical×horizontal) pixels. Each of these 64 (8×8) pixels (one indicated by a group of O marks) is, for example, 8-bit data. In FIG. 7, block data 2 of image signal Sv
1 corresponds to one of the blocks in the frame, and is a matrix of 8 pixels vertically and 8 pixels horizontally. The block data 22 obtained by cosine-transforming this consists of a matrix of horizontal U and vertical V, and both U and V are units of frequency. However, the coefficients after cosine transformation (indicated by a group of 64 ◯ marks, as in the case of block data 21) are usually expressed by the following formula.

In the above formula, f (i, j) is block data 2
1, and c (i, j, u, v)
is a function including cosine parameters that can be expressed as i, j, u, and v, and N is a constant value determined by how to set the range of pixel levels. And on the left side F (u,
v) represents the coefficient after cosine transformation. This coefficient consists of a 0 coefficient that becomes 0 after passing through the quantizer 13 and a significant coefficient that does not become 0. In the block data 22, significant coefficients are shown with hatching. What is clear from looking at this block data 22 is that the cosine transformation of the image signal SV is concentrated on the low frequency side. That is, block data 22
U (frequency increases in the direction of the right arrow) and V (
(frequency increases in the direction of the downward arrow), significant coefficients are concentrated in the upper left half. By the way, the origin (
u=0. v=o) is a DC component. Based on this property, the number of data to be encoded as significant coefficients is drastically reduced. Note that the O coefficient is also sent as data to the transmission path, but the O coefficient
Since the frequency of occurrence of the coefficient is high, the code length to be assigned to it is small, for example 2 bits. On the other hand, since significant coefficients have a code length of 3 bits or more (the less frequently they occur, the longer the bit string is assigned), the fewer the number of significant coefficients, the more convenient it is to improve the efficiency of transmission path use and reduce transmission costs. is good. For example, according to the zigzag scan method, E OB (En
The code d of B 1ock is attached, and the subsequent transmission of the O coefficient is discontinued. On the receiving side, the O coefficient after EOB is reproduced by itself for each block.

[Problem that the invention seeks to solve]

In the above-mentioned conventional image signal processing apparatus, all of the significant coefficients up to the last significant coefficient marked with EOB are encoded and sent to the transmission path without omitting any detail.

However, when the distribution of significant coefficients and O coefficients was investigated, it was statistically found that isolated significant coefficients especially occur among significant coefficients. An isolated significant coefficient is an isolated significant coefficient located far away from further significant coefficients belonging to the lower frequency side. With conventional methods, it doesn't matter whether it's isolated or not.
Anyway, 0 up to the significant coefficient (on the higher frequency side than the last one)
All coefficients and significant coefficients were encoded and sent to the transmission path.

However, even if a coded signal (Sc) is generated based on the coefficients including the last isolated significant coefficient and a considerable number of O coefficients existing before it and transmitted to the receiving side, the amount of information to be transmitted is large. However, there is a problem in that the image quality on the receiving side does not improve that much. Moreover, such a large amount of information to be transmitted cannot be said to fully utilize the effect of band compression by two-dimensional orthogonal transformation.

[Means for solving problems]

FIG. 1 is a block diagram of the principle structure for realizing the system of the present invention. Compared to the general image signal processing device described above (FIG. 6), an isolated significant coefficient remover 31 is added, and isolated significant coefficients are excluded from the encoding target in the entropy coder 14.

[For production]

The image signal processing device 30 shown in FIG. 1 includes an isolated significant coefficient remover 31, and among the significant coefficients subjected to two-dimensional orthogonal transformation in each block, significant coefficients belonging to the higher frequency side and to the lower frequency side are removed. Isolated significant coefficients located far away from the significant coefficients of the next layer to which they belong are removed. And entropy coder 1
In step 4, data that does not include such isolated significant coefficients is encoded. The encoded signal S'c thus obtained
has a higher band compression degree than the encoded signal Sc in FIG.

〔Example〕

FIG. 2 is an explanatory diagram illustrating a first example of encoding according to the present invention. This figure corresponds to the block data 22 after the two-dimensional orthogonal transformation already shown in FIG.

Therefore, the hatched ○ mark is a significant coefficient,
The other O marks are 0 coefficients. This first example shows a mode in which the encoded signal S'c is transmitted by zone scanning, and the isolated significant coefficient 41 belonging to the highest frequency side is set to one of the four corners.
zone (p.

An encoded signal S'c is obtained by two-dimensionally scanning (like a rask scan in a general TV) the inside of a rectangle with q on two sides. Conventionally, regardless of whether the significant coefficient 41 is isolated or not, everything within the zone is encoded as a scanning target. However, in the present invention, the significance coefficient 4
If 1 is determined to be an isolated significant coefficient, it is excluded from the encoding target. To determine whether or not it is an isolated significant coefficient, draw a circle with a predetermined radius r centered around the last significant coefficient (41 in this figure), and if all the coefficients within the circle are 0 coefficients, This is done by making it an isolated significant coefficient.

The isolated significant coefficient remover 31 in FIG.
All the block data shown in Figure 2 is temporarily stored in this RAM, and based on this stored data, the significant coefficient of 1 on the higher frequency side is extracted and its radius is It is determined by pattern matching whether all the stored data in r have O coefficients. The isolated significant coefficient 41 thus determined is the first
It is excluded from encoding by the entropy coder 14 shown in the figure.

FIG. 3 is an explanatory diagram illustrating a second example of encoding according to the present invention, and like FIG. 2, it corresponds to the block data 22 after two-dimensional orthogonal transformation in FIG. 8. This second example shows a mode in which the encoded signal S'c is transmitted by zigzag scanning, and the broken line 52 in the figure represents the zigzag scan route. Conventionally, regardless of whether the significant coefficient 51 is isolated or not,
The last significant coefficient with the EOB mentioned above (51 in this figure)
) were encoded as scan targets. However, in the present invention, if the significant coefficient 51 is determined to be an isolated significant coefficient, it is excluded from the encoding target. To determine whether or not it is an isolated significant coefficient, when the number of consecutive 0 coefficients from the last significant coefficient to the other significant coefficient appearing before it exceeds a predetermined number, the last significant coefficient is isolated. This is done by making it a significant coefficient. For example, if the predetermined number is "7", the significant coefficient 51 in this figure
is the isolated significant coefficient. In other words, the immediately preceding significant coefficient 53
Since there are seven consecutive O coefficients up to this point, significant coefficient 51 becomes an isolated significant coefficient and is excluded from encoding. 8 consecutive O coefficients up to 53, 9 consecutive...
When it becomes ., the significant coefficient 51 naturally becomes an isolated significant coefficient.

In this second example as well, the previously described RAM that temporarily stores all block data is used.

FIG. 4 is a diagram showing another aspect (third example) of the second example of FIG. 3, and shows an example in which the last significant coefficient is not necessarily an isolated significant coefficient. In this figure, the last significant coefficient (significant coefficient on the highest frequency side) to which EOB should be attached is 55 as shown in the figure, and conventionally, up to this significant coefficient 55 was targeted for encoding. On the other hand, according to the present invention, from this significant coefficient 55, there are four consecutive O coefficients of 1-2-3-4 while tracing back the zigzag scan to the immediately preceding significant coefficient 55, and the predetermined O coefficient is The number of consecutive items does not exceed 7. Therefore, the significant coefficient 55 cannot be an isolated significant coefficient. Next, from the significant coefficient 54, the O coefficient is 5 of 1-2-3-4-5 while going back through the zigzag scan and reaching the immediately preceding significant coefficient 51.
They are continuous and do not exceed a predetermined number of consecutive pieces, 7. Therefore, the significant coefficient 54 cannot be an isolated significant coefficient either. Then, if we go back further to the immediately preceding significant coefficient 53, seven consecutive 0 coefficients appear as in the case of Fig. 3, and the significant coefficient 53 at this time appears.
1 is an isolated significant coefficient. In this case, not only the significant coefficient 51 but also the subsequent significant coefficients 54.55 are excluded from the encoding target. Even if these significant coefficients 51, 54, and 55 are excluded, their influence on image signal reproduction is small compared to more significant coefficients on the low frequency side, and the image quality does not deteriorate significantly.

FIG. 5 is a circuit diagram showing a specific example of the isolated significant coefficient remover in the second embodiment. In this figure, quantizer 1
A scan converter 61 is provided to output the data from 3 in the opposite direction of the zigzag scan. That is, for example, scan data is read from coefficient 56 of FIG. 4 in the direction of coefficients 55-54. Eight flip-flops 63-1 and 63-2 hold these in sequence.
...63-7.63-8. The reason why the number of flip-flops is eight is because the predetermined number of consecutive flip-flops mentioned above is set to "7", and if this is "8", one more flip-flop is required. These flip-flops 63-1 to 6
3-7 cooperates with the OR gate 62 to check whether there are seven consecutive 0 coefficients.

The flip-flop 67 has 64 cycles (6 blocks of block data).
If all the coefficients detected through the OR gate 66 are O, the flip-flop 67 outputs "0". However, if a significant coefficient is output to the entropy coder 14, it outputs "1".The OR gate 65 outputs 7 consecutive coefficients when the output of the flip-flop 67 is "0" (all coefficients so far are "0"). The coefficients of are all “0”
'' (the output of the OR gate 62 is 0''), it becomes 0'', and the AND gate 64 causes the flip-flop 63-8 to
Replace the coefficients within with “0”. If the coefficient in 63-8 is a significant coefficient, the targeted isolated significant coefficient is removed.

FIG. 6 is a circuit diagram showing an example of an isolated significant coefficient remover in the third example embodiment. In this figure, the scan converter 61' is the scan converter 61 of FIG.
Unlike the case of , scanning is performed in the forward direction. In other words, scanning is performed from the origin of block data. The flip-flop 68 is reset (set to "L" level output) every 64 cycles (corresponding to 64 pieces of block data). This flip-flop 68 maintains the state of whether there have been seven or more consecutive "0"s, and outputs "1" when there have been no seven or more consecutive "0"s. If there are seven or more consecutive ``0''s, 0'' is output. This 0° output causes the AND gate 69 to be closed thereafter. As a result, 7
If there are consecutive “O”s, all subsequent coefficients are O.
Replace with coefficients. Here, the targeted removal of isolated significant coefficients is performed.

〔Effect of the invention〕

As described above, according to the present invention, by consciously excluding isolated significant coefficients on the high frequency side and not encoding data that does not contribute much to image quality, it is possible to improve transmission efficiency and reduce transmission costs. can be achieved.

[Brief explanation of drawings]

Figure 1 is a block diagram of the principle configuration for realizing the system of the present invention. Figure 2 is an explanatory diagram illustrating a first example of encoding according to the present invention. Figure 3 is an illustration of a second example of encoding according to the present invention. FIG. 4 is a diagram showing another aspect (third example) of the second example in FIG. 6 is a circuit diagram showing a specific example of an isolated significant coefficient remover in the third embodiment; FIG. 7 is a Pronk diagram showing an example of a general image signal processing device; FIG. 8 is a known known example. It is a figure which shows the aspect of the cosine transformation of. 12... Two-dimensional orthogonal transformer, 13... Quantizer, 14... Entrobi coder,
21... Block data of image signal, 22... Block data subjected to two-dimensional orthogonal transformation operation, 30... Image signal processing device, 31... Isolated significant coefficient remover, 41.51... - Isolated significant coefficient, 3v...image signal, Sc, S'c...encoded signal.

Claims (1)

[Claims] 1. Block data (22) in which one frame composed of an image signal (Sv) is divided into a plurality of block data (21), and a two-dimensional orthogonal transformation operation is applied to each block data. In the image signal processing device (30) that encodes the image signal, each of the block data (21)
Among the coefficients forming the block data (22) obtained by applying a two-dimensional orthogonal transformation operation to An image signal encoding method characterized in that isolated significant coefficients (41, 51) are excluded from the encoding target. 2. The block data (22) is obtained by zone scanning.
When encoding a significant coefficient that belongs to a higher frequency side and has all 0 coefficients within a predetermined radius around this significant coefficient, the isolated significant coefficient (4
1) The system according to claim 1. 3. The block data (22
), the number of consecutive 0 coefficients that are significant coefficients on the higher frequency side and that appear before the zigzag scan is a predetermined number. The isolated significant coefficient (5
1) The system according to claim 1.