JP3200517B2 - Image signal encoding method and image signal encoding device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a case where a single image signal is divided into a plurality of pieces in a high-efficiency coding of a television signal and the coding is performed. BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image signal encoding method and an image signal encoding device for removing unnaturalness near a division boundary generated in an image signal obtained by decoding and synthesizing each divided image by performing control.

[0002]

2. Description of the Related Art When one picture signal is divided into a plurality of pieces and coding is performed, each divided picture is regarded as one picture and coding is performed independently. When encoding is performed in a divided manner, parallel processing is possible, and real-time encoding of high-resolution video such as HDTV is easily realized. Also, it becomes possible to use an encoding tool for a standard TV.

FIG. 12 is for explaining an example of a conventional division encoding method. The number of divisions of one image signal is 4
It is assumed that a discrete cosine transform (hereinafter, referred to as DCT) is used.

First, an image input from an input terminal 901 is divided by an image divider 902 into four divided image signals (for example, A, B, C, and D shown in FIG. 4). The divided image signal 903 is encoded by each of the encoding devices 904 to 907. An output 916 of each of the encoding devices 904 to 907 is multiplexed by a multiplexing unit 917 and output to an output terminal 918 as encoded data of an input image. The quantization control of each divided image is performed independently, and each of the encoding devices 904 to 90
7 at the quantization control units 908 to 911.
The quantization step of each divided image is determined based on the occupation states of 2 to 915.

According to such a method, encoding and decoding can be individually performed for each divided image. As a result, when encoding is performed with N division, encoding can be performed in approximately 1 / N of the time when encoding without division.

[0006]

The case where the image signal is encoded without division and the case where the image signal is divided and encoded are compared. It is assumed that the quantization step is determined from the occupation state of the buffer.

FIG. 13A shows a case where an image signal is encoded without division, and FIG. 13B shows a case where an image signal is divided and quantization control is performed independently for each divided image signal.
The change of the quantization step of a certain line is shown.

When encoding an image signal without dividing it,
As the quantization step of the entire image signal is determined by the quantization control unit based on the buffer in the encoding device, FIG.
As shown in (a), the quantization step is a continuous value.

On the other hand, when an image signal is divided and encoded, a buffer is provided for each encoding device for each divided image. In this case, since the quantization step of each divided image is determined independently based on the occupation state of the buffer in the encoding device for each divided image, the occupation state of the buffer changes independently for each divided image encoding device. However, the quantization step near the division boundary may be significantly different from the quantization step near the division boundary of adjacent divided images as shown in FIG. As described above, since the quantization step near the boundary of each divided image may be significantly different, there has been a problem that the boundary of the divided image is detected.

[0010]

SUMMARY OF THE INVENTION The present invention has been made to solve this problem, and employs the following method.

First, when an image signal is divided into N signals and encoding is performed for each of the divided image signals, the quantum signals used in the already-encoded region of the divided image adjacent to the encoding target divided image are used. With reference to the quantization step, the quantization step used near the division boundary of the encoding target divided image is corrected so as to approach the quantization step.

FIG. 1 shows the flow of a quantization process according to the present invention. In the coding of a divided image, a quantization step Q which is a basis of a region to be coded is calculated (S1), and it is determined whether or not the region is near a division boundary (S2). If the region is other than near the division boundary, quantization is performed as it is (S6). If the block is near the division boundary, it is determined whether or not quantization step correction is performed (S3).
Judgment of presence / absence of this quantization step correction (correction area judgment)
Is used as a criterion for determining whether or not the encoding process of the region near the division boundary has been completed in the adjacent divided image. If the encoding process of the region near the division boundary of the adjacent divided image has not been completed yet, quantization is performed without correcting the quantization step Q, and the quantization step Q is the quantization of the region near the division boundary of the adjacent divided image. It is output to the encoding unit of the adjacent divided image for the correction of the conversion step (S4). If the encoding process of the region near the division boundary of the adjacent divided image has already been completed, the quantization is performed by referring to Q ′ of the region near the division boundary of the adjacent divided image output from the encoding unit of the adjacent divided image. Step Q is corrected (S5), and then quantization is performed (S6).

FIGS. 2A and 2B are diagrams for explaining the time difference of the encoding process between the divided images. FIG.
(B) shows a case where the image is divided on the left and right. For example, when encoding is performed from the upper left to the lower right of the image, the area a of the divided image A illustrated in FIG. Therefore, since the quantization step used in the region b is determined before the region a is encoded, the quantization step used in the region b is determined when the quantization step used in the region a is determined. The quantization step is determined to be close.

FIG. 3 is a diagram showing an example of correction of a quantization step used near the division boundary. As described with reference to FIG. 2, since the area a of the divided image A is encoded after the area b of the divided image B, the quantization step of the divided image B is corrected using the quantization step of the divided image B. As a result, the quantization step near the division boundary is approximated as indicated by a circle in FIG.

[0015]

As described above, in the vicinity of the boundary between the divided images, the quantization steps used between the divided images have close values, and the boundaries between the divided images can be made inconspicuous.

[0016]

In this embodiment, as shown in FIG. 4, the number of divisions of one image signal is 4, A, B, C, and D, and DCT is used. Assuming a case where encoding is performed downward and each divided image is processed in parallel, the area where the quantization step is corrected near the division boundary is set to three blocks.

FIGS. 5 to 8 are flow charts of the processing in the embodiment of the present invention. First, the quantization process of the divided image A will be described with reference to FIG. The divided images adjacent to the divided image A are the divided image B and the divided image C. Before the quantization processing of the adjacent region of the divided image A, the quantization processing of the adjacent region of the divided image B and the divided image C is completed. are doing. Therefore, first, a quantization step Q is calculated (S11), and it is determined whether or not the area is near the division boundary (S12).
Alternatively, the quantization step Q is corrected with reference to the quantization step used when the adjacent area of the divided image C is quantized (S13). Thereafter, quantization is performed using the corrected quantization step (S14). In addition, quantization is performed without correcting the quantization step Q in an area determined to be other than the vicinity of the division boundary (S14).

Next, the quantization processing of the divided image B will be described with reference to FIG. The divided image adjacent to the divided image B is
Before the quantization processing of the adjacent area of the divided image B, the quantization processing is completed only for the adjacent area of the divided image D, and the adjacent area of the divided image A is subjected to the quantization processing. Is not finished. Therefore, the quantization step Q
Is calculated (S21), it is determined whether or not the area is near the division boundary (S22). In the area determined to be near the division boundary, it is determined whether the area is adjacent to the divided image A or the area adjacent to the divided image D (S23). Split image A
If it is determined that the region is adjacent to the divided image A, the quantization step Q is output to the encoding unit of the divided image A (S24), and quantization is performed without correcting the quantization step Q (S26).
If it is determined that the region is adjacent to the divided image D, the quantization step Q is corrected by referring to the quantization step used when quantizing the adjacent region of the divided image D (S25). Quantization is performed using the result (S26). In addition, quantization is performed without correcting the quantization step Q in an area determined to be other than the vicinity of the division boundary (S26).

Next, the quantization process of the divided image C will be described with reference to FIG. The divided image adjacent to the divided image C is
Although the divided image A and the divided image D are included, the quantization process is completed only in the adjacent region of the divided image D before the quantization process of the adjacent region of the divided image C, and the adjacent region of the divided image A is quantized. Is not finished. Therefore, the quantization step Q
Is calculated (S31), it is determined whether or not the area is near the division boundary (S32). If the area is determined to be near the division boundary, whether the area is adjacent to the divided image A or the area adjacent to the divided image D is determined. Is determined (S33). When it is determined that the area is adjacent to the divided image A, the quantization step Q is output to the encoding unit of the divided image A (S34), and quantization is performed without correcting the quantization step Q (S36). When it is determined that the area is adjacent to the divided image D, the quantization is performed after correcting the quantization step Q with reference to the quantization step used when quantizing the adjacent area of the divided image D (S35). (S36). In the region determined to be other than the vicinity of the division boundary, quantization is performed without correcting the quantization step Q (S36).

Next, the quantization process of the divided image D will be described with reference to FIG. The divided image adjacent to the divided image D is
Although the divided image B and the divided image C are included, the quantization processing has not been completed for the adjacent region of the divided image B and the divided image C before the quantization processing of the adjacent region of the divided image D. Therefore, a quantization step Q is calculated (S41), and it is determined whether or not the area is near the division boundary (S42). In the area determined to be near the division boundary, the quantization step Q is output to the encoding unit of the divided image B or the divided image C (S43), and quantization is performed without correcting the quantization step Q (S44). . In the area determined to be other than the vicinity of the division boundary, quantization is performed without correcting the quantization step Q (S44).

Next, FIG. 9 shows a block configuration of an embodiment of the present invention. First, an image input from the input terminal 101 is divided by the image divider 102 into four divided image signals 103. The divided image signal 103 is input to each of the divided image encoding devices 104 to 107, and is encoded through discrete cosine transform units 115 to 118, quantization units 119 to 122, code assignment units 123 to 126, and the like.

Here, the quantization steps used in the three blocks near the division boundary are corrected according to the distance from the division boundary. The quantization step to be corrected is determined by the buffers 127-1 of the divided image encoding devices 104-107.
30 based on the occupation state of each of the 30 quantization control units 108 to 111
Use the value determined in.

FIG. 10 is a diagram for explaining a method of correcting the quantization step. As shown in FIG. 10A, the quantization step A ′ _{m, n} of the corrected divided image A is calculated according to, for example, the following equation in the left and right divided regions.

A ′ _{m−2, n} = (3 · A _{m−2, n} + B _{1, n} ) / 4 A ′ _{m−1, n} = (2 · A _{m−1, n} + B _{1, n} ) / 3 A ' _{m, n} = (A _{m, n} + B _{1, n} ) / 2 This corrects the quantization step.

Further, in an area vertically divided as shown in FIG. 10B, the quantization step _{A'm, n} of the corrected divided image A is calculated according to the following equation, for example. A ' _{m, n-2} = (3.A _{m, n-2} + B _{m, 1} ) / 4 A' _{m, n-1} = (2.A _{m, n-1} + B _{m, 1} ) / 3 A ' _{m, n} = ( _{Am, n} + _{Bm, 1} ) / 2 This corrects the quantization step. Where A _{m, n}
Represents a quantization step of the divided image A before correction, and B _{m, n} represents a quantization step used for a block near the division boundary of the adjacent divided image B.

Next, reference will be made to a quantization step used for a block near a division boundary of an adjacent divided image when correcting a quantization step of a block near the division boundary of the encoding target divided image. FIG. 11 shows how to refer to the quantization step near the division boundary between the divided images A to D.

First, in a region where the divided image A and the divided image B are adjacent to each other, the quantization control unit 108 of the divided image A
With reference to the quantization step 113 used for the adjacent block of the divided image B that has already been encoded, the quantization step of three blocks in the divided image A near the division boundary with the divided image B is corrected. In a region where the divided image A and the divided image C are adjacent to each other, the quantization control unit 108 of the divided image A refers to the quantization step 112 used for the adjacent block in which the encoding of the divided image C has already been completed. Then, the quantization steps of the three blocks near the division boundary between the division image A and the division image C are corrected.

In a region where the divided image B and the divided image D are adjacent to each other, the quantization controller 109 of the divided image B
The quantization step 114 used for the adjacent block of the divided image D that has already been encoded is referred to, and the quantization step is corrected to three blocks near the division boundary of the divided image B with the divided image D.

Similarly, in a region where the divided image C and the divided image D are adjacent to each other, the quantization control unit 110 of the divided image C uses the quantization used for the adjacent block for which the encoding of the divided image D has already been completed. Referring to step 114, the quantization step of three blocks near the division boundary between the division image C and the division image D is corrected.

Here, since the encoding process is performed on the adjacent blocks of the divided image D before any of the adjacent divided images, the quantization control unit 111 does not correct the quantization step.

Except for the three blocks near the division boundary of each divided image, the quantization steps used in other divided images are not referred to, and the quantization control units 108 to 111 independently perform quantization control. Is performed, and the quantization step is determined based on the occupation state of each of the buffers 127 to 130.

The coded data 131 of each of the divided image coding devices 104 to 107 is multiplexed by a multiplexing unit 132 and output to an output terminal 133 as coded data of an input image.

In the embodiment described above, the number of divisions is four, but any number of divisions is possible. Further, the division form, the divided image referring to the quantization step, the area setting near the division boundary, the method of correcting the quantization step, the image encoding method, and the like are not limited to the examples described above.

[0034]

As described above, according to the present invention,
When the image signal is divided and encoded, the quantization step used in the region near the division boundary can be set to a value close to each divided image. As a result, unnaturalness in the vicinity of the boundary of image division can be removed.

[Brief description of the drawings]

FIG. 1 is a flowchart of a quantization process according to the present invention.

FIG. 2 is a diagram illustrating a time difference of an encoding process between divided images.

FIG. 3 is a diagram illustrating an example of correction of a quantization step near a division boundary.

FIG. 4 is a diagram illustrating an example of division of an image signal.

FIG. 5 is a flowchart of an embodiment of the present invention.

FIG. 6 is a flowchart of an embodiment of the present invention.

FIG. 7 is a flowchart of an embodiment of the present invention.

FIG. 8 is a flowchart of an embodiment of the present invention.

FIG. 9 is a diagram showing a block configuration of an embodiment of the present invention.

FIG. 10 is a diagram illustrating a method of correcting a quantization step.

FIG. 11 is a diagram illustrating a reference relation of a quantization step near a division boundary between divided images.

FIG. 12 is a diagram illustrating a configuration example of a conventional apparatus for dividing and encoding an image signal.

FIG. 13 is a diagram illustrating a problem of quantization control at the time of division.

[Explanation of symbols]

 Reference Signs List 101 input terminal 102 image divider 103 divided image signal 104 to 107 divided image coding device 108 to 111 quantization control unit 112 quantization step for divided image C 113 quantization step for divided image B 114 quantization step for divided image D 115-118 Discrete cosine transform unit 119-122 Quantization unit 123-124 Code allocation unit 127-130 Buffer 131 Output coded data of each coding device 132 Multiplexing unit 133 Output terminal

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04N ^7/ 24-7/68 H04N 1/41-1/419

Claims (2)

(57) [Claims] An image signal encoding method for dividing an image signal of a current frame into a plurality of image signals with respect to a digital image signal and performing encoding for each of the divided image signals . When encoding the area near the division boundary , the encoding of the area near the division boundary in the adjacent divided image is
If not completed, the encoding of the encoding target divided image
Is performed using the quantization step defined in the area, and the
Quantization step is the amount of area near the division boundary of adjacent divided images
Output to the encoding unit of adjacent divided images for
And the encoding of the region near the division boundary in the adjacent divided image
If it has been completed, it is output by the encoding unit of the adjacent divided image.
See quantization step for correction of the applied quantization step
Then, the quantization scale determined in the encoding target divided image is
Correct the step and use the corrected quantization step
The encoding of the region near the division boundary of the encoding target divided image is performed.
An image signal encoding method characterized by performing . 2. An image signal encoding apparatus comprising: means for dividing an image signal of a current frame into a plurality of image signals with respect to a digital image signal; and divided image encoding means for encoding each divided image signal. in the means for transmitting the quantization step information used in the one divided image encoding means to another adjacent divided image encoding means, the coding target division in each divided image encoding means
Means for calculating a quantization step of an image;
When encoding the area near the image boundary,
Encoding of the area near the division boundary in the split image has been completed.
Then, it is transmitted from the other adjacent divided image encoding means.
Means for correcting the calculated quantization step by referring to the obtained quantization step information .