WO2008067761A1

WO2008067761A1 - Method and apparatus of video coding and decoding

Info

Publication number: WO2008067761A1
Application number: PCT/CN2007/071167
Authority: WO
Inventors: Yannan Wu; Yun He; Jianwen Chen; Xiaozhen Zheng
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2006-12-04
Filing date: 2007-12-04
Publication date: 2008-06-12
Also published as: CN101198052B; CN101198052A

Abstract

A method of video coding includes following steps: use the first prediction mode to code a random accessing entry image in a non-coded video sequence; take the said random accessing entry image as reference image and the fourth prediction mode is used to code n images which follow the said random accessing entry image in the coding order, and n is a pre-establishing positive integer; the said random accessing entry image or/and coded images after the said random accessing entry image are used as reference image to code other non-coding images which follow the said random accessing entry image in the coding order. And a method of video decoding, an apparatus for coding, a apparatus for decoding and an apparatus for coding and decoding. The subject matter is economized code rate than a technology of video random access in prior art.

Description

Video coding method, decoding method and device thereof

The present invention relates to the field of video signal processing, and in particular, to a video encoding method, a decoding method, and a device thereof.

Background of the invention

A good video encoding and decoding method needs to support the random access operation effectively under the premise of ensuring the highest possible compression efficiency, that is, when the image address with the random access point is transmitted to the decoder, the decoder can be correct. The image with random access points and the image after the image are decoded. At present, the prediction modes of video image sequences used for encoding can be classified into the following three types: The first type of prediction is to encode the current pixel by using the encoded pixel data in the same image; For example, the current image is encoded by using chronological order, that is, playing or displaying reference image data in front of the current image; the third type of prediction is reference image data and time after chronological order after the current image. The current image is encoded by reference image data preceding the current image. The bitstream data obtained by the three coding methods has a flag code for identifying the nature of an image, and the code indicates that the graph uses the first type of prediction mode for encoding. The current image referred to above is the image being encoded or decoded. The one image can be a frame image or a field image.

The first type of prediction method has the lowest coding efficiency because it uses only the data in the image. The second type of prediction method uses the reference image data before the current image time to predict the current image, and its compression efficiency. The compression efficiency is higher than that of the first type of prediction method; the third type of prediction method predicts the current image by using both the reference image data after the current image time and the reference image data before the current image time. Its compression efficiency is The highest of the above three encoding methods. However, of the above three types of coding techniques, only the first type can support random access operations.

The existing MPEG-2 video codec standard and the H.264 video codec standard use the above three types of prediction techniques in combination. For convenience of description, the image encoded by the first type of prediction mode is referred to as an A picture, The image B image encoded by the second type of prediction mode, and the image coded by the third type of prediction mode are C pictures. In MPEG-2, a video sequence is divided into a series of image groups, each image group having a picture group header, that is, a description of the image group features, by finding the starting identification code of the image header. To access random access to the group of images. In each group of pictures, the A picture is followed by the picture group header to support random access, and then the B picture or C picture is used as much as possible to improve compression efficiency.

In MPEG-2, when a C picture is introduced in encoding, it is likely that a C picture belonging to the previous picture group in the code stream follows the A picture of the next picture group. When random access occurs, since the image before the current image group corresponding to the random access point cannot be obtained, immediately after the A image of the current image group, the first non-B image subsequent to the A image It is possible that the C picture between the images cannot be decoded normally. To solve this problem, MPEG-2 reserves a picture group enclosing flag in the picture group header to indicate whether the picture group can be randomly accessed.

The method of using the closed flag for video random access operation is as follows:

Determining whether the C picture (if any) after the first A picture after the picture group header in the code stream uses only the fourth type of prediction mode, which utilizes the reference picture data after the current picture in chronological order To predict the current image, or use the first type of prediction mode encoding, if it is, then the closed flag is set to 1, otherwise set to 0. When the closed flag is 1, if random access occurs, even if the image before the current image group is not available, the image in the current image group can still be decoded normally; if the closed flag is 0, the current image is discarded. The group performs a random access and continues to search backward for a new group of images. In H.264, the B picture and the C picture use a multi-reference frame technique, that is, the number of reference pictures at the time of predictive coding is more than one. When a C picture using the third type of prediction mode or a B picture of the second type of prediction mode having multiple reference pictures is introduced into the coding, it is possible to have a C picture which belongs to the A picture before the A picture in the code stream. In the case after the A picture, or the case where the B picture immediately after the A picture is referred to the image before the A picture. When random access occurs, the image before the current A picture may not be available, and the B picture or C picture in which the above occurs may not be decoded normally. To solve this problem, H.264 introduces an instant decoding update image identifier to support random access. If the current image is an instant update image, all the reference images before the current image will be marked as unavailable, that is, in the encoding order, the encoding order is the time sequence of the image encoding process, and the image is in the instant update map. For the image after the image, the image before the image is updated as the reference image when the second type of prediction mode is encoded. By using an instant update image, H.264 can also achieve good random access to video.

Although the above prior art solves the problem of random access of video, it brings a new problem, which is to make the coding efficiency lower. Since the random access method is solved in accordance with the MPEG-2 or H.264 standard, the prediction direction of the C picture or the B picture immediately following the A picture and the number of optional reference pictures are limited. This results in a reduction in the coding efficiency of the set of C pictures or B pictures. Specifically, a group of C pictures after the A picture and before the B picture after the A picture can only perform the fourth type of prediction mode coding, and cannot perform the third type of prediction mode coding; The first B picture following the A picture can only be referenced one picture forward, and the subsequent reference picture of the B picture is also limited by the number. For a video sequence with a large number of random access points, for example, a broadcast video sequence requires an average of 0.5 seconds to have a random access point, and the problem of reduced coding efficiency caused by the existing video random access method is more prominent. Summary of the invention

In view of the above, an object of the present invention is to provide a video encoding method that reduces random access to a video stream while reducing the efficiency of video sequence encoding due to random access techniques. The encoding method includes the following steps:

The random access cut-in image in the uncoded video sequence is coded by the first type of prediction;

n images having a coding order located after the random access cut-in image are encoded as a reference image by using the random access cut-in image as a reference image, and n is a predetermined positive integer;

For the other uncoded pictures whose encoding order is after the random access cut-in image, the randomized access cut-in image or/and the randomized access-cut image after the cut-in image are encoded as reference pictures.

Still another object of the present invention is to provide a decoding method corresponding to the above encoding method, comprising the following steps:

Decoding the random access in the undecoded video sequence into the image;

For the decoding of the image encoded by the fourth type of encoding after the random access cut-in image, n is a predetermined positive integer;

Inserting the random access into the image, and the decoded image encoded by the fourth type of encoding after the random access of the cut image as a decoded reference image, and the decoding order after the random access of the cut image The undecoded image is decoded.

It is also an object of the present invention to provide a video encoding apparatus and decoding apparatus. The encoding device includes the following modules:

a random access judging module, configured to receive an uncoded video sequence, determine whether the received current image has a random access punctured image, and if yes, send the current image and the subsequent image to the random access coding module, otherwise The encoded video sequence is sent to the normal encoding module; a random access coding module, configured to encode the received random access access image and subsequent images according to a random access coding rule, and output the encoded video sequence; a normal coding module, configured to receive the received video The current image and subsequent images are encoded according to normal encoding rules, and the encoded video sequence is output.

The decoding device includes the following modules:

a random access judging module, configured to receive a video sequence, determine whether the received current decoded image is a random access punctured image, and if yes, send the video sequence to a random access decoding module, otherwise send the video sequence to normal decoding Module

a random access decoding module, configured to decode a random access access image and subsequent images in the received video sequence according to a random access decoding rule, and output the decoded video sequence;

The normal encoding module is configured to decode the received video sequence according to a normal decoding rule, and output the decoded video sequence.

The present invention also proposes a codec device composed of the above-described encoding device and decoding device. As can be seen from the above technical solution, at the time of encoding, at least one image after random access to the cut image is encoded by the fourth type of prediction mode; thus, for random access to other images after the image is cut, the image can be used. The randomly accessed punctured image and/or the image encoded by the fourth type of prediction mode described above is encoded as a reference image. Since the random access of the cut-in image and the reference image of the image after it does not include random access to the image before the cut-in image, random access with the random access of the cut-in image as a point of entry can be realized at the time of decoding. Compared with the prior art, the scheme of the present invention does not need to impose a mandatory limitation on the selection of the reference image at the time of decoding, and the code rate is also significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

1 is a code flow chart of an embodiment of the present invention;

2 is a flowchart of decoding according to an embodiment of the present invention;

3 is a block diagram of an encoding apparatus according to an embodiment of the present invention;

Fig. 4 is a block diagram of a decoding apparatus according to an embodiment of the present invention.

Mode for carrying out the invention

The core idea of the present invention is that the image after the random access point is not encoded with the image before the random access point as a reference image, so that random access to the video sequence according to the random access point can be realized. This random access point is the point of entry for random access of video. In the code stream, it can be an image with a random access identifier, which is called a random access cut-in image.

In the embodiment of the present invention, the fourth type of prediction mode is used to assist the random access of the video image, and the fourth type of prediction mode uses the reference image data after the current image in the display order to encode the current image. Hereinafter, for convenience of description and corresponding to the prior art, an image encoded by the first type of prediction mode is referred to as an A image, and an image encoded by the second type of prediction mode is referred to as a B image, and is predicted by the third type. The image obtained by the mode encoding is called a C image, and the image encoded by the fourth type of prediction is called a D image. Among them, the compression efficiency of the A image is low, but the random access is supported; the compression efficiency of the B image and the C image is sequentially improved, but since they need to decode the image located before them as the reference image, the random image is not supported. Access; D images have good compression efficiency and do not require decoding of images that are in sequence before the A image as reference images.

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings. In the first embodiment, if the encoding method of a video sequence is such that there are two C images between the A image and the B image or between the B image and the B image, there is a relationship between the A image and the A image. 4 B images, ie the coding order is as follows:

· · · · · · At the same time, the video sequence can be expressed in the order of time:

A^C.B.C.C^C.C^C^B^C^A^C^ ...... , The time sequence shown below will be referred to as the display order.

The subscript number is the serial number of the image in the display order. At the time of normal encoding, the image c ₁₄ , the image c ₁₅ and the image can be forward referenced to the image. If the image ₆ is set as a random access cut-in image of the segment of video, then according to the inventive arrangement, after the image A ₁₆ is encoded, any C image after the image A ₁₆ and before the image B _{19 is} changed. For encoding in the fourth type of prediction, for example, in this example, the first image whose encoding order is immediately followed by the image A ₁₆ is encoded by the fourth type of prediction, so that the original image C ₁₄ becomes an image. D ₁₄ , whose reference image is image A ₆ . The image D ₁₄ and the image B ₁₄ are identical in display order and encoding order. Then the coding sequence of the video is Α _λ B ₄ C ₂ CB ₇ C ₅ C ₆ 5 ₁₀ C ₈ C ₉ B _l C _{l λ} C _l2 A _l6 D _l4 C ₁₅ B _l9 C ₁₇ C ₁₈

According to the solution of the present invention, the image D ₁₄ can be used as the encoded reference image of the image c ₁₅ and the image β ₁₉ , and the image c ₁₅ and the image cannot use the image before the A ₆ as the encoded reference image. . Upon decoding, when a video segment is set to the random access cut image ₆ appears, decoding the first image ₄₆ a first type prediction mode, then the image of a decoding order after ₆ The image encoded by the fourth type of prediction mode is decoded, and the image is used as a reference image of the first C image and the first B image immediately following the image. According to the above operation to complete the decoding, the video can achieve random access with 4 ₆ as the cut image.

From the specific application of the above solution of the present invention, it can be seen that the coding scheme of the present invention has the following key points:

A. For the random access cut-in image, the first type of prediction mode is used for coding; B. One or more images in the coding order after random access to the cut-in image are encoded in a fourth type of prediction mode, and the reference image is a random access cut-in image;

C. For the other uncoded image whose encoding order is after the random access cut image, the randomized access cut image or/and the random access cut image after the encoded image is used as a reference image. coding. Specifically, the image encoded by the first type of prediction mode and/or the image of the fourth type of coding mode are used as reference images.

As shown in FIG. 1 , the encoding processing flow of the second embodiment of the present invention includes the following steps: Step 101: Determine whether the current image is set to randomly access the cut image, and if yes, execute step 102, otherwise perform processing according to the normal encoding process.

Step 102: encoding the random access cut-in image by using the first type of prediction manner to obtain

A image.

Step 103: The one or more images whose encoding order is located after the random access cut-in image are encoded by the fourth type of prediction mode, and the reference image is the A image to obtain a corresponding D image. In the present embodiment, an image whose encoding order follows the random access cut-in image is encoded in a fourth type of prediction.

Step 104: Perform a second type of prediction mode or a third type of prediction mode coding on the current image by using the encoded image in the coding order after the A image and the A image and before the current image as a reference image. , get B image or C image.

Step 105: Determine whether there is still an uncoded image. If yes, execute step 106. Otherwise, end the encoding process.

Step 106: Determine whether a new random access cut-in image appears, and if yes, go to step 102, otherwise go to step 104.

The corresponding decoding process is shown in Figure 2, including the following steps:

Step 201: If it is determined that the currently decoded image in the video sequence is a random access cut-in image, step 202 is performed; otherwise, processing is performed according to a normal decoding process. Step 202: Decode the A image corresponding to the random access image into the image.

Step 203: Decode the D picture immediately after the random access cut image A picture in the decoding order, and the reference picture is an A picture.

Step 204: Decode the current decoded image by using the image after the random access to the image and before the current decoded image as a reference image.

Step 205: Determine whether there is still an undecoded video image, and if yes, execute step 206, otherwise, end the encoding process.

Step 206: Determine whether a new random access cut-in image appears, and if yes, go to step 202, otherwise go to step 204.

The image encoded by the fourth type of prediction in the second embodiment is the next image randomly accessed in the coding order. In the prior art, the number of C pictures between the A picture and the first B picture after the A picture, or between two adjacent B pictures is usually a fixed value, which is set to m. Any one or more of the m C pictures can be changed to be encoded in the fourth type of prediction mode in the scheme of the present invention to obtain a D picture.

Embodiment 3 of the present invention is a random access coding apparatus and a decoding apparatus for implementing the scheme of the present invention. The random access coding device is as shown in FIG. 3, and includes:

The random access judging module 301 is configured to receive an uncoded video sequence, determine whether the received current image is a random access punctured image, and if yes, send the current image and the subsequent image to the random access encoding module 302, Otherwise, the current image and its subsequent image are sent to the normal encoding module 303.

The random access coding module 302 is configured to encode the received random access hand-in image and subsequent images according to a random access coding rule, and send the encoded video sequence to the storage module 304. The random access coding rule satisfies the following process in principle:

A. The current random access cut-in image is encoded by the first type of prediction mode;

B. In the uncoded video sequence, the coding sequence is in the random access cut-in image The subsequent n images are encoded by the random access cut-in image as a reference image, and encoded by a fourth type of prediction, where n is a predetermined positive integer;

c. encoding, for the encoded sequence, the other uncoded image after the random access of the cut image, the encoded image after the random access cut image or/and the random access cut image as a reference image .

The specific content of the random access encoding rule depends on the specific format of the video sequence. For example, the code stream according to the coding mode of the first embodiment can be coded by referring to the coding mode of the first embodiment of the present invention.

The normal encoding module 303 is configured to encode the received current image and subsequent images according to normal encoding rules, and send the encoded video sequence to the storage module 304. The normal coding rule may be any video sequence coding method that does not need to implement random access, for example, a coding method in a second type of prediction mode or a third type of prediction mode.

The storage module 304 is configured to store the received video sequence and output the stored video sequence.

As shown in FIG. 4, the random access decoding apparatus of the embodiment of the present invention includes:

The random access judging module 401 is configured to receive a video sequence, determine whether the received image is a random access punctured image, and if yes, send the video sequence to the random access decoding module 402, otherwise send the video sequence to normal. Decoding module 403;

The random access decoding module 402 is configured to decode the image corresponding to the random access identifier in the received video sequence and the subsequent image according to a random access decoding rule, and output the decoded video sequence. The random access decoding rule corresponds to the random access coding rule in the above coding apparatus, and follows the following process:

a, decoding random access in the undecoded video sequence into the image;

b, for the decoding sequence, after the random access cut-in image, n images encoded by the fourth type of encoding are decoded, n is a preset positive integer; C. cutting the random access into the image, and/or decoding the decoded image encoded by the fourth type of encoding after the random access of the cut image as a decoded reference image, and cutting the random access in the decoding order The other undecoded pictures after the picture are decoded.

The normal encoding module 403 is configured to decode the received video sequence according to a normal decoding rule, and output the decoded video sequence.

The display module 404 is configured to receive the decoded video sequence from the random access decoding module 402 or the normal encoding module 403, and display the video code sequence as a video image.

It has been shown by a large number of experiments that the scheme of the present invention can significantly save the code rate as compared with the MPEG-2 and H.264 random access methods in the case of also supporting random access.

The above is only the preferred embodiment of the present invention, and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. Within the scope.

Claims

Claim

A video encoding method, comprising the steps of:

And n uncoded images whose encoding order is located after the random access cut-in image are encoded by the random access cut-in image as a reference image according to a fourth type of prediction manner, where n is a preset positive integer;

The encoding method according to claim 1, wherein before the random access hand-in image in the uncoded video sequence is encoded by the first type of prediction, the method further comprises: determining the uncoded video sequence Whether a random access punctured image occurs, and if so, the step of encoding the random access punctured image in the uncoded video sequence using the first type of prediction mode is performed, otherwise the prior art video sequence encoding process is performed.

The encoding method according to claim 1, wherein the n uncoded images are k-th uncoded images whose encoding order is after the random access cut-in image, and k is preset Positive integer.

The encoding method according to any one of claims 1, 2 or 3, wherein said encoded image after said random access cut image or/and said random access cut image is used as The reference image is an image obtained by the first type of prediction and/or an image obtained by the fourth type of prediction as a reference image.

The encoding method according to claim 4, wherein said encoded image after said random access cut image or/and said random access cut image is used as a reference The image is encoded as: using the second type of prediction mode or the third type of prediction mode, encoding the encoded image after the random access cut image or/and the random access cut image as a reference image.

6. A video decoding method, comprising: the following steps:

Decoding the random access in the undecoded video sequence into the image;

Decoding the undecoded image encoded in the fourth type of prediction mode after the random access plunge image, and n is a predetermined positive integer;

Placing the random access into the image, and/or the decoded image encoded by the fourth type of prediction after the random access of the cut image as a decoded reference image, and the random access cut image for the decoding order Other undecoded pictures are then decoded.

The decoding method according to claim 6, wherein the decoding of the random access punctured image in the undecoded video sequence further comprises: determining whether the current image is a random access punctured image, and if so The step of decoding the random access in the undecoded video sequence into the image is performed, otherwise the prior art video sequence decoding process is performed.

The decoding method according to any one of claims 6 or 7, wherein the n uncoded images encoded in the fourth type of encoding are: decoding order after the random access punctured image The kth undecoded image, k is a pre-set positive integer.

The decoding method according to claim 8, wherein the decoding of the other undecoded image after the random access of the image is decoded into: using a second type of prediction method or a third type of prediction In the manner, the decoding sequence is decoded after the random access of the cut-in image.

10. A video encoding apparatus, comprising:

The encoding device according to claim 10, wherein the encoding device further comprises: a storage module, configured to store a video sequence from a random access encoding module or a normal encoding module, and output the stored video sequence .

12. A video decoding device, comprising:

The decoding apparatus according to claim 12, wherein said decoding means further comprises a display module for receiving the decoded video sequence and displaying it as a video image.

A video encoding and decoding apparatus, comprising an encoder and a decoder, wherein the decoder comprises:

a random access judging module, configured to receive an uncoded video sequence, determine whether the received current image has a random access punctured image, and if yes, send the current image and the subsequent image to the random access coding module, otherwise The encoded video sequence is sent to the normal encoding module; the random access encoding module is configured to encode the received random access access image and subsequent images according to a random access encoding rule, and output the encoded video sequence; a normal encoding module, configured to encode the received current image and subsequent images according to a normal encoding rule, and output the encoded video sequence;

The decoder includes:

a random access decoding module, configured to decode the cut-in image and the subsequent image in the received video sequence according to a random access decoding rule, and output the decoded video sequence;