EP0800747A1

EP0800747A1 - Digital video programme multiplexing device

Info

Publication number: EP0800747A1
Application number: EP95942247A
Authority: EP
Inventors: Jean-Michel Masson
Original assignee: Matra Communication SA
Current assignee: Nortel Networks France SAS
Priority date: 1994-12-13
Filing date: 1995-12-12
Publication date: 1997-10-15
Also published as: FR2728129B1; WO1996019081A1; CA2208432A1; AU4350196A; CN1173264A; FR2728129A1

Abstract

A device for transmitting a number of video programmes over a transmission channel having a predetermined throughput capacity, including, for each programme channel supplying a multiplexer, a video encoder having a frame-by-frame programmable encoding throughput. The device comprises, on each programme channel, a frame store (20) with a capacity of at least one frame positioned upstream from the programmable encoder (16), and an auxiliary encoder (22) receiving a series of input frames one frame ahead of the the first encoder. Means (26) for comparing the qualities of the frames encoded by the auxiliary encoders with one another or with a reference value control the encoding throughput to balance the output qualities of the various encoders.

Description

DEVICE FOR DIGITAL MULTIPLEXING OF PROGRAMS

VIDEO

The present invention relates to devices for the real-time transmission of several video programs, each in the form of a binary data stream representing successive frames, on a transmission channel having a determined bit rate capacity, the device comprising, for each program channel supplying a multiplexer attacking the channel, a video coder having a programmable coding rate.

The document EP-A-0 526 201 describes such a video coder, making it possible to adjust, for each group of blocks of pixels of an image, the quantization step so as to maintain the quality of the image at an approximately value constant.

In the rest of the text, the term "weft" should be interpreted in a broad sense. It can designate all of the video data necessary to reconstruct a complete image in progressive scanning, all of the data necessary to reconstruct an even or odd frame in interlaced scanning, or even assistance data making it possible to reconstruct an image with from a previous image. The invention finds a particularly important, although not exclusive, application in transmission devices according to the MPEG2 video standard, which was the subject of recommendation ITU-TH262 (ISO / IEC) 13818-2 published in June. 1994 by the International Telecommunication Union).

The MPEG2 standard provides for the implementation of compression algorithms making it possible to generate modifiable bit rate flows for standard resolution images (appendix D.4 to the recommendation above). This possibility is used in particular when coding a delayed-time film for recording on a storage medium.

Transmission channels have limited capacity. The solution used at present when several programs are multiplexed on the same channel (satellite channel for example) consists in defining a priori, for each program, the maximum bit rate which it must apply to the multiplexer. This solution is not very satisfactory because it misuses the resources of the channel. The richness of the video information transmitted in each of the programs is very variable. At a given instant, it is statistically probable that at least one coder then has at his disposal a bit rate higher than that which is necessary for him, while another coder will have to apply a compression rate reducing the iconographic quality of the 'image, because it has an insufficient flow.

The article by Guha et al "multichannel joint rate control of VBR MPEG encoded video for DBS applications" in IEEE Transactions on Consumer Electronics, Vol. 40, No. 3, August 1994 proposes an algorithm intended to allocate, to different coders supplying the same multiplexer, bit rates which are a function of the complexity of the successive groups of images to be coded and such that the cumulated bit rate is not likely to cause an overflow of a buffer register upstream of the multiplexer. This algorithm does not allow, as such, to carry out assignments in real time, since we do not know the images which will be subsequently received and their richness in information.

The present invention aims in particular to provide a transmission device of the type defined above, responding better than those previously used to the requirements of practice, in particular in that it allows the best use of the capacity of the transmission channel.

To this end, the invention uses the existence of coders making it possible to code each frame at a bit rate different from the previous one. It also uses the fact that we know analyze the information from the coding to determine the iconographic quality of the image obtained.

The invention therefore proposes a device which comprises: - on each program channel, an image memory having a capacity of at least one frame, placed upstream of the encoder, and an auxiliary encoder receiving the successive input frames at the same time as memory;

means for comparing the quality of the frames coded by the auxiliary coders of each channel, with each other or with a reference, and for controlling the coding rate on the different channels to balance the qualities of the outputs of the different coders.

Such a device performs optimal statistical multiplexing, the nature of which is completely different from that of video coding with a quantization step adjusted for each group of blocks to take account of the sensitivity to distortion.

In the particular case of the MPEG2 standard, which provides compressed streams with a bit rate between 2 and 15 Mbits / s in main-main profile, and transmission over a satellite channel, the most frequent case is that of the transmission of five programs on the same channel. In this case, the invention will allow, instead of arbitrarily allocating 8 Mbits / s to each program channel, to adapt the bit rate assigned to it at any time to the richness of the image. . The coded video stream has a suitable bit rate and is then split into packets divided into sequences. The above characteristics as well as others will appear better on reading the following description of a particular embodiment given by way of example. The description refers to the figure which accompanies it and which is a simplified block diagram, showing a particular mode of implementation of the invention, in the case of a device with two coders supplying the same channel. The device, the basic structure of which is shown in FIG. 1, comprises two video program channels 10, and 10 ₂ supplying a multiplexer 12 which attacks modulation and transmission circuits not shown, since they are not concerned by the invention. Each program channel comprises, in addition to the video stream, a binary stream representing the sound and a stream of service information. They can be added to the video stream upstream of the multiplexer 12. In practice, the functions of packetization, multiplexing of the video, sound and service data and of constitution of the queue will be performed in circuits. 14a and 14b which it is not necessary to describe here, because they can be of known type. Each program video channel comprises a programmable encoder 16 with variable bit rate, having a compression rate control input 18. The video stream to be coded is applied to the variable bit rate encoder 16 by means of a memory of image 20 having a capacity at least equal to one frame. Synchronization means, not shown, allow writing into memory 20 as the video stream corresponding to a complete frame arrives, and reading from memory with a delay such that the variable bit rate coder 16 receives the stream video with a frame delay. Each channel also includes an auxiliary coder 22 which receives the video stream at the same time as the memory 20. The auxiliary coder 22 is provided for coding each frame with a fixed bit rate, which will generally be such that the resources allocated to each of the channels are the same. In the case of a satellite channel which can accept a bit rate of 40 Mbits / s, the auxiliary encoder 22 will then be provided for coding each image at the video bit rate compatible with a bit rate on the satellite channel of 8 Mbits / s.

The auxiliary coder 22 is of a type making it possible to analyze the information resulting from coding and to provide an iconographic quality index. Encoders capable of providing such output information are already known. In particular, an encoder of this type is available from the company C-Cube Microsystems, Milpitas, California. It can also be used as a programmable encoder.

Various iconographic quality assessment criteria exist and can be used. Mention may in particular be made of the following. A first possibility consists in analyzing the average, over the whole of a frame, of the quantization step implemented in the algorithm for passing from the time domain to the frequency domain (algorithm for transformation into discrete cosine in general). A given quantization step corresponds to a flow. When the image is complex, the video coder increases the quantization step during the frame for the groups of pixel blocks concerned, so as not to excessively increase the number of bytes representing the image. Analysis of the mean of the final quantization step over a frame makes it possible to know whether the coder would have required a higher bit rate to represent the image satisfactorily and therefore what quality he obtained with the programmed bit rate. In the particular case of MPEG2 coding, which uses intra-frame coded images (I images) coded without reference to other images, predictively coded images (P images) and images coded by bidirectional prediction (B images). , a statistic on the respective number of macroblocks used makes it possible to evaluate the level of activity of an image and to deduce therefrom a quality index for a given compression rate.

Another possibility consists in measuring the signal-to-noise ratio between the image before coding and the image after coding. / 01648

6

The coding device also includes a processor 26 which receives the iconographic quality indications provided by the two auxiliary coders 22, which play the role of quality estimators. The auxiliary coders 22 are provided so that they provide an indication of quality to the processor 26 after a period of time less than the frame repetition period. The processor 26 is programmed so as to compare the quality indices it receives either between them or with a reference index. If for example the iconographic qualities of the coded images are comparable, the processor 26 controls the programmable coders 16 so that they use the same compression ratio.

If, on the other hand, the quality indices show that the iconographic quality of one of the images coded at the nominal bit rate is notably lower than that of the other image coded at the same nominal bit rate, the processor then programs the video coder of the channel having a richness of image requiring a high bit rate for coding by the corresponding coder 16 to be carried out at a bit rate higher than the nominal bit rate. The distribution algorithm can be of the same kind as that described in the article by Guha et al already cited. It can be implemented by cable or software.

The processor 26 and the programmable video coders 16 can in particular be provided to cause coding at a nominal bit rate and at one or two bit rates above and one or two bit rates below the nominal rate, this simple solution being sufficient in most cases.

The processor 26 must simultaneously have the evaluations provided, each over a complete frame, by the coder estimators 22 of all the program channels. To facilitate this comparison, it is desirable that the video streams are synchronized, by means not shown, so that the frames are in temporal coincidence. However, it is possible to give up synchronization tion, provided there is an image memory of increased capacity upstream of each estimator coder.

Various video coding standards impose codings of different natures on successive frames. This is in particular the case of the MPEG 2 standard, already mentioned, whose I images require a volume of information approximately three times greater than P images and approximately nine times greater than B images.

One might think that the problem thus posed can be solved by synchronizing the coders 16 (and the coders 22) so that the comparison is always made between frames of the same kind. With this solution, the volume of information varies from frame to frame within very considerable limits. This would lead to a very high compression rate for the I frames. It is more advantageous to accept that the processor 26 randomly receives frames of different nature at a given time, and to provide the coder estimators so that they take into account the volume of information required by the different types of frame for the same quantification step.

The invention is susceptible of numerous variant embodiments. The number of channels can be greater than or less than five. The auxiliary coders 22 which serve as estimators can have multiple coding rates rather than just one. The different channels can be allocated different resources, which leads to giving the auxiliary encoders 22 different bit rates.

If sufficiently long queues are accepted, the bit rates can be distributed on the basis of an estimate over a determined number of frames, greater than 1. A drawback of this solution is that image memories must be provided upstream of the encoders. It is also possible to ensure such distribution over several frames by storing surpluses or residues of rates at the end of each comparison by the processor 26, and taking them into account on one frame or following frames.

Claims

1. Device for transmitting several video programs each in the form of a binary data stream, each consisting of a video sequence representing successive frames, on a transmission channel having a determined bit rate capacity, comprising, for each program channel supplying a multiplexer (12), a video coder (16) having a programmable controllable coding rate, characterized in that it comprises:

- on each program channel, an image memory (20) having a capacity of at least one frame, placed upstream of the programmable encoder (16), and an auxiliary encoder (22) receiving the input frames successive, with a frame in advance on the programmable encoder, and

- Means (26) for comparing the quality of the frames coded by the auxiliary coders, with each other or with a reference, and for controlling the coding rate to balance the qualities of the outputs of the different coders and use the entire capacity of the channel.

2. Device according to claim 1, characterized in that said auxiliary encoders (22) are provided to provide a quality index to said comparison means.

3. Device according to claim 1 or 2, characterized in that the comparison means comprise a processor (26) providing a selection signal between several coding rates to said programmable coders (16).

4. Device according to claim 1, 2 or 3, caracté¬ ized in that the auxiliary encoders are provided for coding the video streams they receive at a bit rate which is the same fraction of the average bit rate available on the channel.

5. Device according to any one of the preceding claims, intended for transmitting according to the MPEG2 standard, on a satellite channel, characterized in that it comprises five program channels each having an auxiliary encoder provided for a bit rate of 8 Mbi s / s.