KR100373602B1 - Video decoder and corresponding process - Google Patents

Abstract

The decoder comprises first storage means S1 for storing the first coded picture B. The first coded picture may be stored until the picture is decoded at least twice. The result of the decoding can be provided to the reproduction device D. Since the present invention eliminates the need for an output frame buffer for the first decoded picture B, there is less need for memory in the decoder.

Description

VIDEO DECODER AND CORRESPONDING PROCESS}

In order to code a video picture, the so-called MPEG-standard, which applies the principles of transform and predictive coding for so-called hybrid coding, is used. This provides three different coded pictures, different in terms of data compression:

1. I-picture (Intra-coded picture): Coding of I-picture is done independently of other pictures. I-pictures are used to synchronize the transmission of multiple consecutive video pictures and their coding uses local correlation of the data of the I-pictures. The coding is preferably a discrete cosine transform (DCT), followed by quantization, weighting of transform coefficients and entropy coding.

2. P-picture (predicted picture): The coding depends on the temporally preceded I- or P-picture (prediction). The P-picture is motion predicted with respect to the preceding picture (moving compensated prediction). Intra-picture-coding (eg, in the manner described above for I-pictures) is then applied to the temporal prediction error.

3. B-picture (bidirectional predicted picture): Here, temporally motion compensated prediction is made for a temporally preceded P- or I-picture and temporally subsequent P- or I-picture. Also called "moving compensated interpolation". The expression "temporarily" or "preceding" is not about the transmission order of the pictures in the data stream of the coded picture, but on its recording / playing-order. A picture that is the basis of temporal prediction for coding of another picture is also called a "support picture". As in the P-picture, the B-picture is coded in the form of quantized transform coefficients of the difference picture.

In addition, especially for display by television according to the PAL- or NTSC-system, the video picture is not reproduced in so-called frames, but in two consecutive fields ("interlaced format"). One frame includes the entire video picture to be displayed and the entire line of the picture. In contrast, one field contains only half of the lines of one video picture. The so-called "top-field" or first field is the line of all even numbers of the relevant frame, and the "bottom-field" or second field is the line of all radix of the relevant frame.

The MPEG-decoder must not only decode the coded video picture in the correct order, but also to provide the two fields required for this, for example if the display is provided for a PAL- or NTSC-compliant television. However, the transmitted video picture can be frame- or field-coded. In the first case the transmitted video picture is a frame, in the second case a field. The field forms one frame in which two will be displayed.

The MPEG-decoder comprises storage means for storing a previously decoded supporting picture. The storage of the I- or P-picture is essential because it is the basis of the temporal prediction of the P- or B-picture and is necessary for its decoding. For example, if a coded B-picture received by a decoder is to be processed, decoding of the B-picture is done using a support picture previously received and stored by the decoder, which is used for its bidirectional prediction.

It is also known to save the memory location by having the support picture necessary for decoding the B-picture not decoded, but stored in the decoder in coded form. However, this method has the disadvantage that the amount of data to be processed by the decoder is significantly increased. Since the same I- and P-pictures are generally used to decode multiple B-pictures, this often needs to be decoded. However, it is important to provide block base coding in accordance with MPEG, in which case the prediction range used for prediction is often beyond the block limit in the supporting picture. Therefore, four blocks of two coded and stored supporting pictures must be decoded for decoding a block consisting of B-pictures. Therefore, eight times the amount of data must be processed compared with the case of storing the decoded supporting picture. However, this eight-fold amount of data must be processed in the same unit time, which can be used for decoding of predicted and coded pictures as usual. In other words, the demand for the performance of the decoder is significantly larger than in the case of a decoded and stored support picture.

Conventional MPEG-decoder is provided with storage means in which decoded B-pictures are stored ("output-frame buffer"). The storage means is necessary in the case of a frame-coded B-picture in a conventional decoder, because as described above, two fields must be formed from the frame by the decoder. The fields must in turn be applied to the output of the decoder in order to display the frame in interlace format. The output frame buffer is necessary for other reasons in the case of field-coded B-pictures: in many applications (eg, to adjust different picture frequencies between the recording system and the playback system) two fields are displayed for display of the frame. In addition, the first field must be repeated once again after the display of the second field.

The present invention relates to a video decoder and a decoding method.

1, 2 and 4 are different embodiments of the present invention,

3 shows a configuration of a conventional MPEG-decoder.

It is an object of the present invention to provide a video decoder having a low memory need.

This object is achieved according to the invention by a decoder according to claim 1 and a decoding method according to claim 7.

According to the invention, the first picture is not decoded (as in the prior art) but stored in the decoder in coded form. At least twice the decoding is done for the generation of the fields necessary for reproduction. After each decoding, the decoding result is output to the output of the decoder, so that it can be reproduced (eg by a television). No output frame buffer as in the prior art is needed. For storage of the coded first picture, as in the prior art, a smaller memory location is needed than the decoded picture is stored. The invention is not limited to the case of an MPEG-decoder. The present invention is concerned in all cases where multiple times of information must be generated from the first picture for reproduction. Unlike the prior art, the stored and coded first picture is newly decoded each time (e.g., two fields from one frame) to generate different information (e.g., the picture must only be partially and incompletely decoded each time). ). In the prior art the picture is first fully decoded, stored and then output several times (e.g. output of two fields of a decoded frame). The present invention can also be applied to a decoding method in which no temporal prediction is made and no supporting picture is provided.

An improvement of the invention relates to a video decoder according to the MPEG-standard. Here the first picture is temporally predicted for the second picture, but it is not necessary for the decoding of another picture (ie it does not need to be decoded for decoding another picture since it is not used for temporal prediction of another picture). . This first picture may for example be a p- or B-picture.

Unlike the prior art in which coded support pictures (ie, I- and / or P-pictures) are stored, the present invention has the advantage that a significantly smaller amount of data is decoded. The reason is as follows: In the above-described prior art, the coded and stored I- and P-pictures, which are based on their temporal prediction before the decoding of the B-pictures, are first decoded. The decoding of the I- and P-pictures must be done again for each B-picture to be decoded. This multiplies the data to be decoded by the decoder.

In contrast, in the present invention, in the worst case, that is, when the frame-coded B-picture is considered as the first picture, the first decoding and the second field for generating the second field for generating the first field according to the present invention. If two decoding is made from the coded frame, the amount of data to be decoded can be doubled. However, preferably the doubling of the amount of data to be decoded can be avoided, since it can be limited to the decoding of lines which are usually necessary for the field to be generated to obtain one field from one coded frame. to be. This applies for example for motion compensation, but also in most cases of inverse quantization and inverse DCT.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

A coded second picture I, a coded third picture P, and a coded first picture B are sequentially applied to an input In of the conventional video picture decoder shown in FIG. This order does not match the recording order or reproduction order, but the order in which data is transmitted, resulting in a coding scheme. It is assumed that the first image B is a B-picture, the second image I is an I-picture and the third image P is a P-picture.

The coded pictures that are directly applied are each temporarily stored in the input buffer 1 and then processed in an entropy decoder 2, inverse quantization and inverse DCT (Dicrete Cosine Transformation) units 3, where each picture is estimated in time. In the case it is compensated by the movement compensation means 4.

Post-processing unit 5 is connected in front of the output of the decoder, which unit-as assumed in the present case-displays the frame-coded pictures I, P, B in fields I _T,. I _B or B _T , B _B or P _T , P _B ) and output in interlace format. The second picture I and the third picture P, which are essential as support pictures for decoding of the first picture B, are decoded and stored in the second storage means S2, while the decoded first picture B Is stored in the output-frame buffer S for subsequent decomposition into two fields B _T , B _B.

The pictures in the Out, which are then decoded by the decoder, can be provided to the reproduction device D for reproduction.

In the prior art, the supporting pictures I, P are stored in a coded state and decoded for decoding the B-picture predicted in time for the supporting picture.

1 shows a first embodiment of the present invention. Like the decoder of FIG. 3 already known in the prior art, the decoder according to the invention also has an input buffer 1, an entropy decoder 2, an inverse quantization and inverse DCT unit 3, a motion compensation means 4 and a post processing unit. (5) (Post Processing) and second storage means S2 for storing the decoded second picture I and the third picture P, which are used as support pictures for subsequent decoding. In this embodiment, the second picture I is assumed to be an I-picture, and the third picture P is assumed to be a P-picture that is predicted in time with respect to the second picture I.

The input In of the decoder includes a coded second picture I, a coded third picture P and a first picture B (in this embodiment with respect to the preceding two supporting pictures I, P). Temporally bidirectionally predicted B-pictures) are applied in turn. The second picture I and the third picture P are decoded in a known manner and stored in the second storage means S2, while the first picture B is first coded in the first storage means S1. Is stored in the form.

The output of the decoder is provided with the fields I _T , I _B , B _T , B _B , P _T , P _{B of the} I-picture (I), B-picture (B) and P-picture (P). The manner provided is the way in which the fields are formed by a decoder in the above order in an interlaced format, for example for subsequent display on a television. The transmission of pictures I, P, B coded in an order different from the order required for reproduction (at the output of the decoder) is made according to the basic principle of bidirectional prediction and is known in the prior art.

The frame coded picture of the present embodiment, that is, the first picture B that represents the frame and is stored in the first storage means S1, uses the support pictures I and P to respectively use two fields B _T and B _B. It is decoded once to form. In some cases, for example, three decodings may be required, in which case, once again, the second field B _B is formed and then once again necessary to reproduce the video picture to which the first field B _T is transmitted. 1 shows that the first field B _T is transmitted second to Out by In-Clamp-Setting.

As can be seen in FIG. 1, further storage means for storing the decoded first picture B or the fields B _T , B _B formed therebetween between the decoding steps 1 to 4 and the output of the decoder. Since there is no such field, the field can be provided directly to the Out without being temporarily stored. Thus, as shown in Fig. 3, the output frame buffer S of the prior art is omitted.

According to the MPEG-standard, it is common for pictures to be decomposed into so-called macro blocks (which are, for example, six blocks of 8x8 pixels) and transmitted. As a result of the transmission, when decoding, so-called macro block lines are first formed instead of individual picture lines-these lines are, for example, groups of 16 picture lines each. The output from the decoder for subsequent display should be done in line fashion. In order to convert the " block " format into a " line " format, a small temporary storage device must be provided in the place where the output frame buffer S (Fig. 3) is disposed in the prior art for the application of the present invention. . Unlike the prior art, the temporary storage device is used only for recording one macro block line among decoded macro block lines (for example, 16 picture lines), respectively, and a field or frame to be decoded (for example, 288 or 576). Is not used to write (picture lines), which is much smaller than the output-frame buffer. After the entire macro block line has been decoded and stored in the temporary storage device, the output of the line system necessary for reproduction can be made. In order to output the decoded picture lines without interruption, it is important to design the temporary storage device to a size large enough to record two macro block lines (for example, 32 picture lines as a whole). After one macro block line is decoded, the macro block line can be output in a picture line manner, while another macro block line is also decoded at the same time, and the formed data is overwritten to the preceding macro block line. are not continuously overwritten.

Since the first storage means S1 may preferably be part of the input buffer 1, the input buffer reaches the input In of the decoder while the picture is being stored in addition to the first picture B to be stored. It should be designed to be larger in size than the prior art so that an image can be recorded. Unlike the prior art (FIG. 3), the output-frame buffer S in which the decoded first picture B is stored is not essential. The first storage means S1 can be designed with a smaller size than the output frame buffer S of the prior art (including the temporary storage device already described in the preceding paragraph, which is required in some cases).

The formation of the fields B _T , B _B from the first frame B is preferably done in such a way that in two cases the entire first picture B is not decoded and only the lines necessary for the field are decoded. Thereby, the amount of data to be decoded can be significantly reduced.

As an alternative to the illustrated embodiment, the supporting pictures I, P in addition to the first picture B are not decoded, but can also be coded and stored. In such a case, much less memory is needed overall inside the decoder, which increases the decoding cost per unit time.

FIG. 2 is a second embodiment of the present invention in which the first picture B (which is also a B-picture in this embodiment) is not a frame but is additionally transmitted unlike the frame-coded B-picture of FIG. The first field B _T , which forms a field-coded frame together with the second field B _B , is different from the first embodiment of FIG. 1. The two fields B _T , B _B are B-pictures and are predicted in time with respect to the support pictures I, P.

In the decoder shown in FIG. 2, the first field B _T is coded and stored in the first storage means S1 because, as already mentioned, the second field B _{B is} connected to the output Out. This is because it may be necessary to output the first field B _T again after the output. For this purpose, the first received picture I (I-picture) is decoded and stored in the second storage means S2 and decomposed into two fields I _T , I _B and output in the above order. (Out) is provided. Next, the third image P applied to the input In is also decoded and stored in the second storage means S2 as well. The coded first field B _T , which is then received at the input In, is stored in the first storage means S1 and at the same time the second image I and the third stored in the second storage means S2. Decoded with the help of the picture P and provided to the Out. Then, the same as the second field B _B until it is stored in the first storage means S1. Thereafter, the first field B _T stored in the first storage means S1 is decoded and output again, and the third image P stored in the second storage means S2 is then filled with the fields P _T,. It is decomposed into P _B ) and output.

In a block method such as MPEG, in order to be able to output individual picture lines instead of decoded macro block lines, a relatively small temporary storage device for recording two macro block lines includes a motion compensation means 4 and a post processing unit. (5) Should be placed between.

In an alternative embodiment of the invention a P-picture may be provided as the second picture I rather than an I-picture, since B-pictures, as is generally known, may This is because it is also predicted with respect to P-pictures. The same applies to the case where the two supporting pictures to be stored in the first storage means are I-pictures.

Also, the second picture I (as an I-picture or a P-picture) may be stored in the second storage means S2 as the only supporting picture (in this case more than in the embodiment of FIGS. 1 and 2). The first picture B may be a pre-predicted P-picture which is coded in relation to the second picture I but which is not used for prediction (which of course is in the MPEG-standard). Does not correspond).

Of course, the coded pictures I and P assumed to be frames in FIGS. 1 and 2 are either fields or coded as fields, but may also be stored as frames in the second storage means.

4 shows another embodiment of the present invention. Although the present embodiment deals with the MPEG-decoder, the present invention is not limited thereto. In FIG. 4, two B-pictures, that is, a first picture B and a fourth picture B ', are sequentially applied to the input In in addition to the two supporting pictures I and P. All pictures I, P, B, and B 'are frame coded. It is assumed to be a frame. In this embodiment, two B-pictures B and B 'are stored in the first storage means S1. By decoding the two pictures B, B 'several times, for example, the fields displayed at the output Out can be formed in the given order. In this case, the first picture B is first decoded twice to form two respective fields, and then the fourth picture B 'is decoded twice, and then the first picture B is again decoded two times. Is decoded once. Other temporal orders are possible, depending on how the playback by the playback device D should take place.

In figure 4 a temporary storage device Z, which has already been mentioned several times, is shown between the movement compensation means 4 and the aftertreatment unit 5. Temporary storage is essential only in a block coding method and can accommodate two macro block lines in an MPEG-decoder. The temporary storage device is thus much smaller than the output frame buffer S according to FIG. 3. If the coding or decoding is not done in a block manner but in a line manner, the temporary storage device Z is not necessary. In such a case, each decoded picture can be output in a direct line manner.

The decoder according to the invention can be a component part of a corresponding encoder which is the place to carry out the motion evaluation for the coding of the video picture to be transmitted. The MPEG encoder basically also includes a decoder for predicting time.

The present invention can be very preferably applied to all video pictures which have no basis for temporally predicting other pictures. The present invention is not only the first picture B, which is assumed to be B-picture and predicted in both directions in time, as shown in the two embodiments of FIGS. 1 and 2 (in addition to the P-picture that is predicted in one direction in time). It also relates to pictures that are not temporally predicted, for example I-pictures.

Claims (8)

First storage means S1 for storing the compression coded first picture B and For first decoding the stored first picture B and then outputting the decoding result to the output OUT for playback using the playback device D and For subsequent second decoding of the stored first picture B in time, and then outputting the decoding result to the output OUT. A video decoder comprising decoding devices 2, 3, 4. The method of claim 1, The coded first picture B is predicted temporally in relation to the second picture I, but is not used to temporally predict another coded picture, The video decoder comprises second storage means S2 for storing the second picture I, The decoding device decodes the stored first picture (B) using the stored second picture (I). The method according to claim 1 or 2, The first picture B is a frame, The decoding device is used for generating a first field (B _T ) by a first decoding and a second field (B _B ) by a second decoding from a first picture (B). The method of claim 3, wherein And the decoding device is used for generating and outputting the first field (B _T ) to the output again by the third decoding of the stored first picture (B) after the second decoding. The method according to claim 1 or 2, The first coded picture B is the first field B _T , The decoding device is configured by successive decoding Stored first field B _T , Coded second field B _B and A video decoder, which in turn is used to generate the stored first field B _T in turn. The method according to claim 1 or 2, A first storage means S1 is used for storing the compression coded fourth picture B 'in addition to the first picture B, The video decoder being used to decode the stored fourth picture likewise several times and to provide the decoding result to the output OUT. The coded first picture B is stored, The stored first picture B is decoded and the decoding result is provided to the output OUT of the decoder for reproduction by the reproduction device D, A method of decoding a compressed coded first picture (B), characterized in that the stored first picture (B) is decoded twice and the decoding result is provided to the output (OUT). The method of claim 7, wherein The coded first picture B is predicted temporally in relation to the second picture I, but is not used to temporally predict another coded picture, The second picture I is stored, Decoding of the first picture (B) is carried out using the stored second picture (I).