Video decoding method. RU patent 2508608.

FIELD: radio engineering, communication.

SUBSTANCE: video decoding method, which is based on fine granular scalability (FGS), involves obtaining image identification information which indicates the reference image of the current block, and decoding the current block using the reference image. The basic level reference image and the corresponding reference image of a higher quality level have the same image identification information, wherein the image identification information includes at least one of frame number information indicating the decoding order, image number information for image identification and reference index information. The problem which, when performing video decoding, is caused by transmission error can therefore be minimised.

EFFECT: efficient video decoding.

4 cl, 9 dwg

The technical field to which the invention relates

The present invention relates to a method for encoding the video signal.

Prior art

The communications environment is currently undergoing changes due to development of broadband convergence (, BcN) for the provision of such services as the integrated wired and wireless network, the combination of broadcast and communications network or convergence, according to Internet-Protocol (IP convergence) using the network, the corresponding Internet Protocol (IP network). The tendency to change the network connection will accelerate in the future. Due to changes in the environment due, end devices (terminals), used in a variety of communications environments, are subjected to various changes, and also changed processing capabilities possessed by the target device. Accordingly, to ensure the video signals are optimized for various communications environments and different terminal devices, you should create a lot of video signals issued in the target device. To provide for optimized video in terminal devices, one video source must include the number of combinations of variables such as the number of transmitted frames per second, resolution and number of bits per pixel, etc. that imposed a heavy burden on the providers of content (content).

In view of the foregoing, providers of content encode the input video signal receiving compressed high-flow of bits (bit rate) video, decode the compressed video data in an input video signal when the target device is accepted request for video information, and encode the original data, receiving video data, suitable to the abilities of the target device for processing of video information, before issuing the video data in the target device. However, this transcoding (for this combination, the decoding and encoding) it is necessary to carry out the process of coding, decoding and encoding, and therefore there is no time delay in the issuance of the video signal to the target device. Consequently, an additional device with sophisticated hardware and algorithms.

To solve this problem, proposed scalable (MVK SVC). This method can provide video signals by encoding the video signal into a sequence of images with the highest image quality and decoding only part (partial sequence of images, selected in an interruptible mode from the General sequence of frames) a sequence of images. The sequence of images that have been codified by the ERM, may reduce the size of video information through the use of spatial scalability or loss in image quality through the use of scalability against the «signal - to-noise ratio (SNR SNR) with low bitrates. The sequence of images that have a small screen size and/or a small number of frames per second is called a basic level, and the sequence of images that have a large screen size and/or a large number of frames per second is called the level of higher quality or improved level.

Although you can submit a video, but the quality of the image by receiving and processing of the parts of the sequence of images encoded scalable manner, as described above, the image quality is much worse, because it reduces the bit rate. One solution to this problem is to ensure the auxiliary sequence of images with low bitrates, for example, the sequence of images that have a small screen size and/or a small number of frames per second. This helper image sequence is called the superior quality (improved level).

The MVK, if part of the bit-stream level of high quality is lost when decoding a particular image, which includes the level of high quality, which is encoded and transmitted, decoder decodes this image using a partially lost bitstream to the level of high quality. Accordingly, the image quality of the original video signals differ from each other. In particular, when the image is suffering from the above problem is the sample images necessary to decode the other images, as well as a key image, the problem is getting worse.

Brief summary of the invention

Accordingly, the present invention is devoted to the method of decoding video, which essentially removes one or more of the problems arising from the limitations and disadvantages of relevant prior art.

The present invention, proposed to solve, is a method of decoding, allowing minimization problem that when decoding the video signal due to a transmission error.

Another object of the present invention, proposed to solve, is to develop ways to manage the buffer decoded images in accordance with the method of decoding, which helps to minimize the problem that when decoding the video signal due to a transmission error.

The task of the present invention can be solved by developing a method of decoding video-based scalability with precise detail (, FGS), consisting in that the information they receive is the identification of the image, indicating the reference image of the current block, and decode the current block using a reference image, the reference image of the base level and the corresponding reference image level of high quality have the same identification information of the image.

List of figures

Accompanying drawings, which are designed to provide better understanding of the invention, illustrate specific embodiments of the invention and, together with a description serve to explain the principle of the invention.

The drawings:

figure 1 presents the notional amounts block diagram of the system scalable encoding video information that is subject to the present invention;

figure 2 shows that illustrates the structure of images used in the method of decoding, and a way to manage the buffer decoded images in accordance with the variation of the implementation of this the invention;

figure 3 presents a block diagram of the sequence of operations, illustrating the way to decode video in compliance with the embodiment of the present invention;

figure 4 presents a block diagram of the sequence of operations, illustrating the way to decode video in compliance with another embodiment of the present invention;

figure 5 shows that illustrates the structure of a buffer decoded images in accordance with the embodiment of the present invention;

on fig.6-6b are presented to illustrate the reordering of reference rooms reference image in accordance with the embodiment of the present invention;

figure 7 shows that illustrates contents of the buffer decoded images, corresponding to the code numbers of the management operations of memory management (); and

on Fig.8 shows that illustrates the structure of a buffer decoded images in accordance with another embodiment of the present invention.

The best way of carrying out the invention

Now turn to a detailed consideration of the preferred options for the implementation of the present invention, which are illustrated in the attached drawings. Wherever possible, the same position will be used on all drawings for the designations identical or similar parts.

In addition, although the terms used in the present invention, are selected from well-known and commonly used terms, some of the terms mentioned in the description of the invention, are selected by the applicant at its discretion as to the details of their semantic content are presented in the relevant parts of the outlined here describe. In addition, you must understand that this invention is not just literally describes the commonly used terms, and corresponds to the meaning content of each term, inherent in him.

Figure 1 presents the notional amounts block diagram of the system scalable encode video which to apply the present invention.

The system is scalable encode video includes encoder 2 and decoder 10. Encoder 2 includes 4 encoder basic level (BE), encoder 6 level of superior quality (EE) and the multiplexer 8 (Mux). Decoder 10 includes demultiplexer 12 (deMux), decoder 14 of the basic level (BD) and decoder 16 level of superior quality (ED). Coder 4 basic level compresses the video input signal X(n) and generates a stream of bits of base level. Coder level 6 higher quality compresses the video input signal X(n) and generates a stream of bits level of high quality using the video input signal X(n) and information generated by the encoder 4 basic level and multiplexer 8 generates scalable stream of bits using a stream of bits of the base level and the bitstream level of high quality. Generated scalable stream of bits transmitted to the decoder 10 on a certain channel, and passed scalable bit stream is divided into a stream of bits level of high quality and a stream of bits of base level 12 decoder 10. Decoder 14 of the basic level takes a stream of bits of the base level and decode this stream of bits of the base level with getting video output Xb(n), and the decoder 16-level higher quality takes a stream of bits level of high quality and decode the bit stream level of high quality with getting video output Xe(n). Video output Xb(n) has a lower resolution and lower image quality than video output Xe(n).

In the structure of images formed a group of images (GI, GOP) hierarchical structure In the images, and the images that make up GUY, are divided into key image and image. One image is from the reference image QB baseline and reference image QE level of high quality. The key image is an image that has the lowest level of temporary and image is an image except for the key image.

In this structure in the current picture possible to use at least one image from images having the time levels which are lower than temporary level of the current image, or equal, as the reference image. For example, when the first image 22 is the I-image-coded internal prediction, the first image 22 encoded without reference image, the second image 24 coded using the first image 22 as the reference image, the third image 26 encoded using the first and second images 22 and 24 as reference images, the fourth image 28 and 30 are encoded using the first and the third image 22 and 26 or second and third images 24 and 26 as reference images, the fifth image, 32, 34, 36 and 38 are encoded using the first and fourth images 22 and 28, the third and fourth images 26 and 28 or the second and fourth images 24 and 30 as reference images. The images are encoded in the following order: first image 22, the second image is 24, the third picture 26, the fourth image 28 and 30 and in the fifth image, 32, 34, 36 and 38. The first and second images 22 and 24 are the key images, and images from the third to the fifth- 26, 28, 30, 32, 34, 36 and 38 are nonkey images. The key image includes the reference image to the baseline and reference image level of high quality and presented in the form of the first image, 22 and the second image 24 figure 2. When image is used as the reference image, refers to the reference image level higher quality nonkey image. The reference image level higher quality nonkey image indicates that the reference image basic level key image has been restored and is represented by the dotted line as images from the third to the fifth- 26, 28, 30, 32, 34, 36 and 38 shown in figure 2.

Figure 3 presents a block diagram of the sequence of operations, illustrating the way to decode video in compliance with the embodiment of the present invention;

First, the decoder receives the encoded video signal on a given channel. This video get (310) information identification of the image, indicating the reference image of the current block. Information identifying the picture may be, for example, information reference index or information rooms image inherent in the reference image. The current block decode (320) using a reference image. The reference image of the base level and the corresponding reference image level of high quality can have the same identification information of the image. Information identification of the image may be information of the reference index or information rooms image inherent in the reference image. The reference image level of high quality can have better image quality than the reference image base level.

When the image including the current block, is a key image, reference image of the current block may be a reference image of the base level or corresponding sample images level of high quality. The reference image of the current block can be obtained from the list of reference images on the basis of authentication information. Current block can be decoded using this reference image. Even in this case, the reference image of the base level and the corresponding reference image level of high quality can have the same identification information of the image. The reference image level of high quality can have better image quality than the reference image base level.

For example, when the current image, as shown in figure 2, is the second picture 24, and P is the depiction of a reference image of the second image 24 is the first image of 22, and the second image 24 is a key image. Accordingly, the second image is 24, which is the current image, you can decode, turning to the master image baseline or reference image level higher quality first image 22. The reference image to the baseline and reference image level of high quality can have the same information to identify the image, and this information is identification of the image may be information of the reference index or information rooms image inherent in the reference image. In another example, when the current image is the fourth picture 28, reference image fourth image 24 is the first and third images 22 and 26, and the fourth picture 28 matches image. In this case, the reference image fourth image 28 includes reference image of the base level of the first image, 22, reference image level higher quality first image 22 or reference image level of high quality third image 26. The reference image level of high quality third image 26 indicates that the reference image of the base level of the third image 26 is already restored.

Figure 4 presents a block diagram of the sequence of operations, illustrating the way to decode video in compliance with another embodiment of the present invention.

First, the first reference image can be obtained (410) from the list of reference images. Here the first reference image is associated with information identifying the picture, and this information identification of the image may be information of the reference index or information rooms image inherent in the reference image. For example, the first reference image in the list of reference images can be obtained using the information of the reference index. The current block decode (420) with the first reference image. The first reference image and the corresponding second reference image can have the same information to identify the image, and the first reference image and the second reference image differ other identifying information. For example, reference image 40 of the base level and the reference image 42 level of higher quality, with the same information to identify the image, you can do differ from each other by marking the reference image 40 basic level as «basic understanding», carried out in the labeling process decoded reference images. The first reference image can belong to the core level, and the second reference image can belong to the level of high quality. Level of high quality can have the image quality is higher than the quality of the image at the basic level. When the image including the current block, is a key image, reference image of the current block may be the first sample images or corresponding second sample images. Current block can be decoded using this reference image.

As shown in figure 3 and 4, when the reference image is the first image of 22 or second image 24, which is a key image, you can provide a reference image baseline or reference image level of high quality for the current image. Alternatively, you can ensure that the first reference image or second reference image for the current image. Accordingly, reference image can be stored in the buffer of the decoded images. In accordance with the method of decoding the video signal according to the present invention, the manner of appointment of a reference number reference image in the buffer of the decoded images, or way of buffering buffer decoded images, such as the way of excluding the reference image, shall be defined anew. Way buffering buffer decoded images in accordance with the present invention will be described in detail with reference to Fig. 5 and 6.

When the first image 22, which is a key image, as well as sample images, as shown in figure 2, is decoded and stored in the buffer of the decoded images, reference image 40 of the base level and the reference image 42 level of high quality, who constitute the first picture 22, saved with the number «0» image. In the second image 24, which is a key image, as well as sample images, the same the first image 22, reference image 44 baseline and reference image 46 level of high quality, who constitute the first picture 22, saved with the number «1» image. In the third picture 26, which is a non-key image, as well as sample images, reference image 48 level of high quality reference image 26 saved with the number «2» of the image. The fifth image, 32, 34, 36 and 38 are not buffered decoded images, except for the cases when the image is not displayed immediately after decoding, because the fifth images are not reference images.

When the current image is a key image and refers to the reference image, having the number «0» image, as the reference image 40 of the base level and the reference image 42 level of high quality reference image 22 have the same information identifying the picture, which represents the number «0» image, reference image 40 of the base level and the reference image 42 level of higher quality, with the number «0» image, must differ from each other. Accordingly, the present invention can use other identification information. For example, reference image 40 of the base level and the reference image 42 level of higher quality, with the number «0» image, differ from each other by marking the reference image 40 basic level as «basic understanding», carried out in the labeling process decoded reference images.

In another embodiment, information identifying the picture, typical for the images stored in the buffer of the decoded images varies depending on what image is decoded in the buffer of the decoded images. Therefore, the first reference image in the list of reference images can be obtained from the list of reference images, and current block can be decoded using this first reference image. The first reference image and the corresponding second reference image have the same information to identify the image, and these first reference image and the second reference image may differ in other identifying information. For example, the first reference image and the second reference image with the same number of images, you can do differ from each other by marking the first reference image or second reference image as a «basic understanding», carried out in the labeling process decoded reference images. The first reference image can belong to the core level, and the second reference image can belong to the level of high quality. In this case, the reference image and the second reference image, you can do differ from each other by marking the first reference image as a «basic understanding».

In another embodiment of the present invention list of reference images can be initialized using a reference image decoding of the current block using a reference image, as described with references 3 and 4. Initialized list of reference images can be reordered. When the list of reference images removed, reference image can be obtained from the reordered list of reference images Reordering the list of reference images will be described with references to fig.6 and 6b.

On fig.6-6b are presented to illustrate the reordering of reference numbers the reference image in accordance with the embodiment of the present invention.

Reordering the reference numbers of the reference represents the command to reassign the smallest reference image that has the greatest correlation with the current image in the list of reference images, and effective management of sample images when the image close in time to the current image, they have less correlation with the current image, than the correlation image that has been removed in time from the current image, when decoding the current image. The reordering reference numbers will be described in detail with links to fig.6 and 6b.

On fig.6 shows the structure of images in a state where the first, second, third and fourth image 22, 24, 26, 28 and 30 decoded and stored in the buffer of the decoded images. When it is desired to decode the fifth picture 34, list 0 reference image that contains the images stored in the buffer of the decoded images, as shown in fig.6b. Images are given reference numbers when the cyclic shift of a list 0 reference images from the fifth image 34 in a counterclockwise direction, and the images are given reference numbers when the cyclic shift of list 1 reference images from the fifth image 34 in a clockwise direction. However, when the first image 22, which was removed in time from the fifth image 34, has a higher correlation with the current image, than the correlation of the fourth image 28, which is close in time to the current image in the list, 0 reference image reference number 0 is remapped the first image of 22, and the reference number 1 is remapped to the fourth image 28, which leads to reordering the list of reference images. When the second image is 24, which was removed in time from the fifth image 34, has a higher correlation with the current image, than the correlation of the fourth image 30, which is close in time to the current image, in list 1 of the reference image and the reference number 1 is remapped the second picture 24, and the reference number 2 is remapped to the fourth image 30, which leads to reordering the list of reference images.

In one embodiment of the present invention, as described above, the reordering of the reference numbers when the reference image is the key image of the reference image to the baseline and reference image level of high quality treated as a single image, which is remapped reference number and, when the reference image is non-key image, is remapped reference number assigned to a reference image of the level of higher quality, which leads to reordering the list of reference images.

As another option to the implementation of the present invention, in the initialization list reference images the same way as when reordering of reference numbers. For example, to receive information identification of the image, indicating the reference image of the current block, and read the reference image stored in the buffer reference images, on the basis of this information, identification of the image. Current block decode using the read out the reference image. The reference image of the base level and the appropriate reference image level of high quality can have the same information to identify the image, and this information is identification of the image includes information of the reference index or information rooms image inherent in the reference image. When the reference image is a key image, reference image of the base level and the appropriate reference image level of high quality treated as a single image, which can be assigned a reference number. When the reference image is non-key image, the reference number can be assigned to a reference image level of high quality. Even when the reference image should no longer be a reference and is erased from the buffer decoded images using the method of the moving window, if the reference image is the key image of the reference image of the base level and the appropriate reference image level of high quality treated as a single image, and the reference image can be erased from the buffer. If the reference image is non-key image, the reference image level of high quality can be erased from the buffer. In this case, the image is erased using the method of «first-come - first-served basis (FIFO), also called the way of a simple queue, and the first erased the image that is first stored in the buffer.

First, when the code number 0, it indicates that the memory management completed. When the code number equal to 1, it indicates that a short reference image marked not sample images. When the code number is 2, it indicates that a short reference image marked not sample images. When the code number equal to 3, it indicates that a short reference image marked not long-term sample images and moved to the long term memory. When the code number is 4, it indicates that the decision on the amount of long-term memory. When the code number is set to 5, it indicates that all sample images are not marked sample images, and contents of the buffer is reset. When the code number equal to 6, this indicates that the current image is labeled a long-term sample images and moved to the long term memory.

Operation, corresponding to a code number , is as relative to the reference image base level, and with respect to the reference image level of high quality, when the current image is a key image, and is relative to the reference image level higher quality reference image, when the current image is non-key image. That is, when the reference image to the baseline and reference image level of high quality is set to the same number identification of the image, reordering the reference number, the initialization list reference images, erase the reference image and buffer management using can be carried out in block image that has the same number as identification of the image. Number identification of the image includes information of the reference index or information rooms image inherent in the reference image.

In another embodiment of the present invention, information identifying the picture is appointed in the block image stored in the buffer of the decoded images, regardless of whether the reference image key image or non-key image. That is, even when the reference image is a key image, a reference image of the base level and the reference image level higher quality reference image is assigned a different identity information of the image. When the reference image is non-key image, reference image of the base level and the reference image level higher quality stored in the buffer of the decoded images, and so one piece of information identification of the image is assigned a reference image level of high quality.

On Fig.8 shows that illustrates the structure of a buffer decoded images in accordance with another embodiment of the present invention.

When the first image 22 (shown in figure 2), which is a key image, as well as sample images, decoded and stored in the buffer of the decoded images, the room 0 image is assigned a reference image 40 of the base level of the first image, 22, a-number-1 image is assigned a reference image 42 level of high quality first image 22. In the second image 24, which is a key image, as well as sample images, the same the first image 22, number 2 image is assigned a reference image 44 basic level of the second image 24, number 3 image is assigned a reference image 46 level of high quality second image 24. In the third picture 26, which is a non-key image, as well as sample images, because the reference image level higher quality is stored in the buffer of the decoded images, number 4 image is assigned a reference image 48 level of high quality. Since the fifth image, 32, 34, 36 and 38 are image, the fifth image is not stored in the buffer of the decoded images, except when the fifth images do not appear directly after decoding.

In another embodiment of the present invention, when reordering of the reference numbers when the current image is a key image, a reference number is remapped only a reference to the image base level, and when the current image is non-key image, a reference number is remapped only reference image level of high quality.

In another embodiment of the present invention, in the initialization list reference images, and buffering buffer using , if the current image is a key image, and the reference image base level, and the reference image level of high quality are the initialization and memory management operations using even when the reference numbers of these images are different, and if the current image is non-key image, the reference image level higher quality reference image undergoes initialization and memory management operations using .

As shown in Fig.8, for example, the command 62 , on which the reference image 50 moves from short-term memory to long term memory, derived from the current image. When the current image is the key image of the short-term memory to long term memory are moved and the reference image base level, and the reference image level higher quality, and when the current image is non-key image, reference image 58 level of high quality reference image is exposed to command 64 .

When the reference to the reference image is no longer required because of what it is erased from the buffer decoded images method of the moving window, if the current image is the key image of the reference image to the baseline and reference image level higher quality erased from the clipboard, even when the reference numbers of these images are different. However, if the current image is non-key image, the reference image level higher quality reference image is erased from the buffer. In this case, the image is erased using the method of «first-come - first-served basis (FIFO), the first erased the image that is first stored in the buffer.

For specialists in the art it will be obvious that within the scope of the claims of the invention can make various modifications and changes of the present invention. Therefore, it is assumed that the present invention covers modification and changes to this invention, provided that they are within the claims of the accompanying claims and cash equivalents.

Industrial applicability

As described above, in accordance with the present invention, when decoded key image, reference image baseline or reference image level higher quality reference image are issued as the reference image, and when decoded image, reference image level higher quality reference image is issued as a reference image. Therefore, you can minimize the problem that when decoding the video signal due to a transmission error. When the video decode using the above-described method of decoding, enables efficient video decoding using a new method of buffering buffer decoded images, which is optimized with the help of the method of decoding.

1. Way to decode video in containing: decoding the level of the basic quality; decoding level of high quality on the basis of the level of the basic quality; and the marking of the image level of the basic quality and image level of higher quality as the reference image, as image level of the basic quality and picture a level of high quality have the same information to identify the image, and at the stage of marking an image level of the basic quality is marked as other information presentation of identification of the image includes at least one of the information frame number that indicates the procedure decoding, information rooms image for image identification and information of the reference index.

2. The method according to claim 1, wherein image as a baseline quality and picture a level of high quality, which have the same information identifying the picture, have the least time resolution.

3. The method according to claim 1, wherein the layer image of high quality has reduced the size of the sampling rate, compared with a picture of the level of the basic quality.