WO2001033861A1

WO2001033861A1 - Mpeg encoding method for logo insertion

Info

Publication number: WO2001033861A1
Application number: PCT/GB2000/004262
Authority: WO
Inventors: Micheal James Knee
Original assignee: Snell & Wilcox Limited
Priority date: 1999-11-05
Filing date: 2000-11-06
Publication date: 2001-05-10
Also published as: GB9926311D0; AU1162701A; GB2356507A

Abstract

When MPEG encoding a video signal, a set of macroblocks is defined to form the site for downstream insertion of a logo. The encoder forces the coding of the set of macroblocks into an intra-coding mode. It also inhibits the use, in coding of any other macroblock, of a motion vector directed to any of the set of macroblocks. The aggregate bit rate over the set of macroblocks is controlled so as to simplify logo insertion or to maintain picture quality in the logo region.

Description

MPEG ENCODING METHOD FOR LOGO INSERTION

This invention relates to compression encoding of video and images and in the most important example relates to MPEG-2 encoding of digital television.

There is a common requirement to insert a logo or other fixed element into a video signal. This is a relatively straightforward procedure with base band video but is considerably more complex if the video signal is received in compressed form. One approach to the problem is to decode the incoming MPEG-2 signal, insert the logo conventionally into the uncompressed video signal and then re-encode. If the decoding and encoding processes are independent, there is the risk of introducing additional coding artifacts into the signal. This risk can be removed if the decoder is arranged to pass to the encoder, an appropriate set of coding decisions. Re-use by an encoder of the decisions taken by a decoder can result in effectively lossless decoding/encoding. This technology, as shown for example in EP 0 765 576 and EP 0 91 3 058. Whilst this technology is effective, there are instances which a technically less complex solution is required.

An alternative approach is to decode only that spatial region of the video signal which is to receive the logo or other element. Because of the use of motion vectors in MPEG-2, it is necessary to decode and re-encode not only the macroblocks comprising the spatial region in which the logo is to be inserted, but also all those other macroblocks which are linked by motion vectors. This remains a relatively complex processing operation. Moreover, the complexity of the processing which is required will vary with the nature of the picture material and particularly with the number of macroblocks which require decoding and re-encoding.

It is frequently the case that a centrally produced video feed is distributed to a number of locations, at each of which a distinctive logo is inserted before local distribution of the video signal. The problem of inserting a logo without introducing coding artefacts is therefore faced at many different locations, on the same video signal.

It is an object of this invention to provide a system solution to the problem of inserting a logo without introducing coding artefacts. Accordingly, the present invention consists in one aspect in a method of motion compensated compression encoding which operates on macroblocks of a picture, comprising the steps of identifying a set of macroblocks defining a site for downstream insertion of a logo or like element, and inhibiting the use, in the coding of any other macroblock, of a motion vector directed to any one of said set of macroblocks.

Suitably, the method comprises the further step of inhibiting the use, in the coding of each macroblock in said set of macroblocks, of a motion vector directed to any macroblock outside said set of macroblocks. Conveniently, the method comprises the further step of forcing the coding of each macroblock in said set of macroblocks into an intra-coding mode.

Advantageously, the method comprises the further steps of selecting a target aggregate bit rate for said set of macroblocks to provide a desired picture quality at the logo or like element and manipulating the coding of each macro block in said set of macroblocks, adding stuffing where appropriate, so as to meet the target aggregate bit rate for the set of macroblocks.

In another aspect, the present invention consists in apparatus for compression encoding, comprising a coding decision generator serving to generate coding decisions which include the allocation of motion vectors respectively to macroblocks of a picture; a site locator serving to identify a set of macroblocks defining a site for downstream insertion of a logo or like element and a coding decision modifier serving to modify the coding decisions in response to the identified set of macroblocks to avoid the use, in the coding of any other macroblock, of a motion vector directed to any one of said set of macroblocks. In yet a further aspect, the present invention consists in a system comprising a motion compensated compression encoder which operates on macroblocks of a picture, and a downstream logo inserter which operates on the encoded signal to insert into the picture a logo or like element, wherein there is defined at the encoder a picture site for the downstream insertion, the site comprising an identified set of macroblocks, wherein the encoder operates to avoid the use, in the coding of any macroblock outside said set of macroblocks of a motion vector directed to any one of said set of macroblocks and wherein the logo insertor operates on the encoded signal to insert a logo or like element at the picture site.

Typically, the system comprises a plurality of downstream logo inserters operating on the same encoded signal to insert at a common picture site respective, different logos or like elements. In one form of the invention, the logo inserter operates to replace said defined set of macroblocks with pre-encoded macroblocks representative of said logo or like element.

Advantageously, the encoder operates to encode said set of macroblocks with an aggregate bit rate for set of macroblocks which is equal to the aggregate bit rate of said pre-encoded macroblocks.

In another form of the invention the logo inserter operates to decode said defined set of macroblocks, to mix the decoded macroblocks with said logo or like element and to encode said logo mixed macroblocks. Advantageously, the system further comprises a quality selector serving to select a target aggregate bit rate for said set of macroblocks to provide a desired picture quality for said logo mixed macroblocks and wherein the encoder operates so as to vary aggregate bit rate for the set of macroblocks in response to the selected target. The invention will now be described by way of example with reference to the accompanying drawings in which:- Figure 1 is a block diagram of a system according to the present invention; and

Figure 2 is a block diagram of an MPEG-2 encoder according to the present invention. Referring initially to Figure 1 , an MPEG-2 encoder 102 receives a video signal input together with a control signal from a "select logo region" block 1 04. This enables a region of the picture to be defined (typically near a corner or edge of the picture) into which a logo will optionally be inserted. This region will comprise a number of macroblocks and preferably consists of a number of whole slices in the MPEG-2 bitstream. In one example, the site comprises the central 1 5 macroblocks of the 45 macroblocks in the picture width. The length of a slice is then defined as one third of the picture width. In the picture height, the site may typically occupy rows 3, 4 and 5 of the 36 macroblock rows. The encoder 1 02 is so arranged that the macroblocks in the selected area are encoded in intra-mode and that no predictions are taken from this area in coding the remainder of the picture. This can be achieved in a variety of ways when coding a macroblock outside the site. The techniques employed may vary from one motion measurement method to the next. Candidate vectors which point to a macroblock inside the site may, for example, be marked "illegal" in the same sense that a motion vector pointing outside the picture would be marked as illegal. Vector refinement processes may be adjusted to keep or to take any "offending" vector out of the selected area. Block matching algorithms may be revised so as to exclude the site area. Where a vector has already been selected, it can be replaced by a close but "non- offending" vector or replaced by the zero vector. Alternatively, the macroblock concerned may be converted to an intra-coded macroblock. The compressed MPEG-2 bitstream, prepared in accordance with this invention, is received by one or a number of logo inserters as shown at 1 06. The logo inserter comprises a splitter 1 08 which identifies the macroblocks in the region which has been defined for subsequent logo insertion. These are passed to a local decoder 1 10. Decoded video is combined in video mixer 1 1 2 with the output of logo generator 1 1 4. The resulting video signal is encoded at local encoder 1 1 6.

The macroblocks outside the logo region are passed directly from the splitter 1 08 to a combiner 1 1 8 in which they are re-combined with the output from local encoder 1 1 6.

In this way, a logo is inserted into the MPEG bitstream with a minimum and constant amount of processing. Outside the prepared area, the processing is transparent. If the logo is the same size as the prepared area and is not transparent, it will be possible to proceed in a simpler manner. In this case, the macroblocks in the prepared area are simply replaced by a pre- encoded logo bitstream.

It will be necessary to pay careful attention to the relative number of bits in the prepared area and in the inserted logo. In the case where a pre-encoded logo directly replaces the prepared macroblocks, the encoder 1 02 will be arranged to ensure through quantiser control and the addition, if necessary, of stuffing, that the aggregate number of bits of the prepared macroblocks equals that of the pre-encoded logo. Alternatively, the logo generator can be arranged to insert stuffing to make up a shortfall of bits.

In the more general case, the encoder 1 02 can be arranged through quantiser control or the use of stuffing, to ensure that there are sufficient bits employed in the prepared area so that the output of the local encoder 1 1 6, producing the same number of bits, offers sufficient visual quality. It will generally be regarded as important that the logo or other fixed element is re-produced on a display with a resolution that does not vary with the nature of the picture material. It will be understood that, without this feature of the present invention, the number of bits available for encoding of a logo would depend upon the number of bits employed in the upstream encoding of the corresponding region of the picture and thus, in turn upon the relative difficulty in encoding the picture and specific regions without the picture.

Turning to Figure 2, there is shown an example of MPEG-2 encoder 1 02. This encoder employs the technology disclosed in the above published patent applications to which reference is again directed.

The incoming video signal is received by a coding decision block 202. This analyses the video signal and makes the necessary MPEG-2 coding decisions. These include coding modes, allocation of motion vectors and quantiser levels. The coding decision block 202 has two outputs: a video output 204 and an information bus output 206. A so- called "dim" coder 208 operates on the coding decisions carried without the information bus signal to encode the video signal 204 and output an MPEG-2 bitstream. In accordance with the present invention, the information bus passes through an information bus processor 21 0 which modifies the coding decisions, as appropriate, to ensure that the macroblocks within a selected area are intra-coded, that no use is made of these macroblocks in prediction coding of the remaining macroblocks of the picture and, preferably, that the defined area is encoded with the appropriate number of bits. For this purpose, the information bus processor 210 will receive control inputs specifying the location of the logo and the relative quality with which the logo is to be displayed. The way that is ensured that no use is made of macroblocks within the site area when coding other macroblocks will vary. In one example, a motion vector pointing to the selected area can be replaced by the closest vector which points outside the area, or by the zero vector. Alternatively, a macroblock having a vector pointing to the selected area can be converted to intra-coding mode.

Whilst the size and position of a logo in a system comprising one encoder and multiple downstream logo inserters will usually remain fixed, applications will arise for the dynamic variation of the location and perhaps the size of the logo site. It may be appropriate, for example, to have editorial control over the logo location to ensure that it masks a less important region of the picture. In this case the location (and if necessary the size) of the logo site can be signaled to the logo inserter as user data or otherwise as a flag in the encoder bitstream. In the same way the number of bits utilised in encoding the macroblocks in the logo site can be signaled. Figure 2 shows the "location" and "quality" signals being taken to the dim coder 208 for inclusion in the encoded bitstream as user data.

It should be understood that this invention has been described by way of examples only and that a wide variety of further modifications are possible without departing from the scope of the invention. Thus, whilst forcing the macroblocks in the site selected for the logo insertion into intra-coding mode is convenient and simplifies the decoding and re- encoding described with reference to Figure 1 , it is not absolutely essential. Particularly where macroblocks in the site are replaced by pre- encoded macroblocks it will be sufficient to ensure that no macroblock outside the selected region utilises a vector pointing inside the selected region. Similarly, whilst the example of a logo has been taken, the invention has application to the insertion of other elements such as text information, both fixed and varying.

Claims

1 . A method of motion compensated compression encoding which operates on macroblocks of a picture, comprising the steps of identifying a set of macroblocks defining a site for downstream insertion of a logo or like element, and inhibiting the use, in the coding of any other macroblock, of a motion vector directed to any one of said set of macroblocks.

2. A method according to Claim 1 , comprising the further step of inhibiting the use, in the coding of each macroblock in said set of macroblocks, of a motion vector directed to any macroblock outside said set of macroblocks.

3. A method according to Claim 1 , comprising the further step of forcing the coding of each macroblock in said set of macroblocks into an intra-coding mode.

4. A method according to Claim 1 , comprising the further steps of selecting a target aggregate bit rate for said set of macroblocks to provide a desired picture quality at the logo or like element and manipulating the coding of each macro block in said set of macroblocks, adding stuffing where appropriate, so as to meet the target aggregate bit rate for the set of macroblocks.

5. Apparatus for compression encoding, comprising a coding decision generator serving to generate coding decisions which include the allocation of motion vectors respectively to macroblocks of a picture; a site locator serving to identify a set of macroblocks defining a site for downstream insertion of a logo or like element and a coding decision modifier serving to modify the coding decisions in response to the identified set of macroblocks to avoid the use, in the coding of any other macroblock, of a motion vector directed to any one of said set of macroblocks.

6. Apparatus according to Claim 5, wherein the coding decision modifier further serves to avoid the use, in the coding of each macroblock in said set of macroblocks, of a motion vector directed to any macroblock outside said set of macroblocks.

7. Apparatus according to Claim 5, wherein the coding decision generator serves to select for each macroblock between inter-picture and intra-picture coding modes and wherein the coding decision modifier further serves to force the coding of each macroblock in said set of macroblocks into an intra-picture coding mode.

8. Apparatus according to Claim 5, further comprising a quality selector serving to select a target aggregate bit rate for said set of macroblocks to provide a desired picture quality at the logo or like element and wherein the coding decision modifier serves to modify the coding decisions so as to vary the aggregate bit rate for the set of macroblocks in response to the selected target.

9. A system comprising a motion compensated compression encoder which operates on macroblocks of a picture, and a downstream logo inserter which operates on the encoded signal to insert into the picture a logo or like element, wherein there is defined at the encoder a picture site for the downstream insertion, the site comprising an identified set of macroblocks, wherein the encoder operates to avoid the use, in the coding of any macroblock outside said set of macroblocks of a motion vector directed to any one of said set of macroblocks and wherein the logo insertor operates on the encoded signal to insert a logo or like element at the picture site.

1 0. A system according to Claim 9, comprising a plurality of downstream logo inserters operating on the same encoded signal to insert at a common picture site respective, different logos or like elements.

1 1 . A system according to Claim 9, wherein the logo inserter operates to replace said defined set of macroblocks with pre-encoded macroblocks representative of said logo or like element.

1 2. A system according to Claim 1 1 , wherein the encoder operates to encode said set of macroblocks with an aggregate bit rate for set of macroblocks which is equal to the aggregate bit rate of said pre-encoded macroblocks.

1 3. A system according to Claim 9, wherein the logo inserter operates to decode said defined set of macroblocks, to mix the decoded macroblocks with said logo or like element and to encode said logo mixed macroblocks.

1 4. A system according to Claim 1 3, further comprising a quality selector serving to select a target aggregate bit rate for said set of macroblocks to provide a desired picture quality for said logo mixed macroblocks and wherein the encoder operates so as to vary aggregate bit rate for the set of macroblocks in response to the selected target.

1 5. A system according to Claim 9, wherein the encoded signal includes a flag representative of the size and location of the picture site for logo insertion.

1 6. A system according to Claim 1 3, wherein the encoded signal includes a flag representative of the said target bit rate.