 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to methods of and apparatus for concealing errors in
digital television signals.
2. Description of the Prior Art
In a digital video tape recorder (DVTR), an incoming television signal to
be recorded is sampled, the samples are pulse code modulation coded into
digital form, the resulting digital data signals are recorded on a
magnetic tape and subsequently reproduced, the reproduced digital data
signals are decoded, and the decoded signals are used to form an analog
signal corresponding to the original television signal. If errors occur in
the handling of the digital signals, for example due to noise or tape
drop-out occurring in the DVTR, the reproduced digital signals are
corrupted and then the re-formed analog television signal does not
correspond exactly to the original analog television signal, and a
resulting television picture is degraded.
Errors are particularly liable to occur when the DVTR is operating in a
special reproduction mode, that is to say in one of the modes which
include still, slow motion, reverse motion and the so-called shuttle mode
which comprises the whole range of speeds from fast forward to fast
reverse.
Such errors and their detrimental effects can be reduced by correction and
concealment. Correction involves the production and use of additional data
signals purely for the purposes of error detection and correction, these
additional data signals otherwise being redundant. While correction
provides good results, it cannot generally be used as the sole means of
dealing with errors, because a comprehensive error correction capability
would require an excessive amount of additional data which might overload
the data handling paths or raise data rate to an unacceptable level.
Concealment comprises the replacement of corrupted data signals by
available uncorrupted data signals and by data signals generated using
available uncorrupted data signals. Concealment relies largely for
accuracy on the strong correlation that exists in a television signal.
In concealment, the data signal which is used for replacement of a
corrupted data signal can be derived spatially or temporally. Spatial
concealment involves the generation of a sample value to replace a sample
value which the earlier-applied error correction process has failed to
correct, by using available sample values in the same field or frame as
the error sample value. To give a very simple example, a concealment
sample value can be generated by averaging the two adjacent sample values
in the same horizontal scan line. Commonly, however, more sophisticated
methods of generating concealment sample values are used, and one such
method is disclosed in our UK Pat. No. 2 084 432. Temporal replacement
involves the replacement of an error sample value by the sample value from
the corresponding sample position in the preceding frame.
Both spatial concealment and temporal replacement have their advantages and
disadvantages. Because usually there is little movement in a television
picture from frame to frame, temporal replacement more usually gives a
better result. However, when movement occurs in the picture, temporal
replacement results in a form of picture degradation known as trailing
dots, which occur at sample positions in the picture where temporal
replacement has been used and hence the picture information has not been
updated. Where movement occurs in the picture, and most particularly where
there is a lot of movement, such as occurs for example when a television
camera is being panned, spatial concealment gives a better dynamic
response and hence a picture which is much more acceptable to the human
eye. With spatial concealment, however, the overall quality of the picture
is reduced.
SUMMARY OF THE INVENTION
One object of the present invention is to provide a method of concealing
errors in a digital television signal in which the problem of trailing
dots is reduced.
Another object of the present invention is to provide a method of
concealing errors in a digital television signal in which a decision is
made in dependence on the picture and error content whether or not to
effect temporal error concealment.
Another object of the present invention is to provide apparatus for
concealing errors in a digital television signal and comprising means to
analyse the movement and error content in the picture to determine whether
or not to effect temporal error concealment.
According to the present invention there is provided a method of concealing
errors in a digital television signal, the method comprising the steps of:
deriving a current sample value and a first group of sample values located
spatially close to said current sample value in the same frame of said
television signal;
deriving from the preceding frame of said television signal a delayed
sample value corresponding in position to said current sample value, and a
second group of sample values respectively corresponding in position to
the sample values of said first group of sample values;
checking said current sample value for error;
checking said delayed sample value for error;
checking pairs of sample values from said first and second groups of sample
values and corresponding in position in said current and preceding frames
of said television signal for errors and to derive a measure of the
difference in the levels of the two sample values in each said pair,
combining the results of these checks to provide an output signal which is
a measure of the number of said errors and of the sum of said differences;
checking said output signal against a predetermined threshold level
corresponding to an acceptably small number of errors and a relatively low
degree of movement in the part of a television picture corresponding to
said groups of sample values; and
if said current sample value is in error, said delayed sample value is not
in error and said output signal is below said predetermined threshold
level, replacing said current sample value by said delayed sample value
for supply to an output, and, in all other cases, supplying said current
sample value to said output.
According to the present invention there is also provided apparatus for
concealing errors in a digital television signal, the apparatus
comprising:
first delay means for deriving a current sample value and a first group of
sample values located spatially close to said current sample value in the
same frame of said television signal;
second delay means for deriving from the preceding frame of said television
signal a delayed sample value corresponding in position to said current
sample value, and a second group of sample values respectively
corresponding in position to the sample values of said first group of
sample values;
means for checking said current sample value for error;
means for checking said delayed sample value for error;
means for checking pairs of sample values from said first and second groups
of sample values and corresponding in position in said current and
preceding frames of said television signal for errors and for deriving a
measure of the difference in the levels of the two sample values in each
said pair, and for combining the results of these checks to provide an
output signal which is a measure of the number of said errors and of the
sum of said differences;
threshold means for checking said output signal against a predetermined
threshold level corresponding to an acceptably small number of errors and
a relatively low degree of movement in the part of a television picture
corresponding to said groups of sample values; and
a selector operative, if said sample value is in error, said delayed sample
value is not in error and said output signal is below said predetermined
threshold level, for replacing said current sample value by said delayed
sample value for supply to an output, and, in all other cases, for
supplying said current sample value to said output.
The above, and other objects, features and advantages of this invention
will be apparent from the following detailed description of illustrative
embodiments which is to be read in connection with the accompanying
drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a very much simplified block diagram of a digital television
reproducing apparatus;
FIG. 2 shows sample arrays in corresponding positions in two successive
frames of a television picture; and
FIG. 3 shows in more detailed block diagrammatic form part of the apparatus
of FIG. 1.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
Referring first to FIG. 1, reproducing apparatus for a digital television
signal recorded on a magnetic tape comprises a DVTR 1, an error correction
arrangement 2, an error concealment arrangement 3, and an output terminal
4 to which the reproduced digital television signal is supplied after
error correction and concealment. The error concealment arrangement 3
comprises a temporal error concealment arrangement 5 and a spatial error
concealment arrangement 6. The error correction arrangement 2 operates in
known manner to effect error correction so far as possible on 8-bit
digital data words reproduced by the DVTR 1 from the tape and representing
successive sample values of the television signal. Each of the output data
words supplied by the error correction arrangement 2 has associated with
it an additional bit forming an error flag which is set to a predetermined
one of the two possible values if the data word still has an error, and is
otherwise set to the other value. The sample values corresponding to the
data words flagged as being in error are concealed by the error
concealment arrangement 3.
As discussed above, temporal replacement generally gives a better result,
in particular for the sample values corresponding to those parts of the
corresponding television picture, which may amount to the whole of the
corresponding television picture, where there is little or no movement.
However, where there is significant movement, spatial concealment gives a
better dynamic response, which is generally more visually acceptable. The
temporal error concealment arrangement 5 therefore operates not only to
affect temporal replacement, but also to judge sample value by sample
value, whether temporal replacement is appropriate. If it is not
appropriate, then spatial error concealment, effected by the spatial error
concealment arrangement 6 is next selected. However, there will be
situations, in particular where there is a very high error rate such as
may arise in special reproduction modes such as the shuttle mode, where
there are simply not enough valid sample values available for any form of
spatial error concealment to give a satisfactory result. In such cases,
temporal concealment is reverted to as the final option, as a temporally
incorrect sample value, even although it may give rise to a trailing dot,
is more acceptable in the reproduced picture than a totally erroneous
sample value.
The present invention is particularly concerned with the form and operation
of the temporal error concealment arrangement 5, and more particularly
with the way in which the primary decision is made whether to conceal an
error samaple value temporally or spatially.
Referring to FIG. 2, this shows an array of nine luminance sample positions
in a frame, the sample positions comprising a central sample position
which is the current sample position under examination and where an error
sample value E0 may exist, and the eight most closely adjacent sample
positions, three in the preceding horizontal scan line, the two adjacent
ones in the same horizontal scan line, and three in the succeeding
horizontal scan line. The sample values at these eight sample positions
will be referred to as S1 to S8, respectively. FIG. 2 also shows the nine
corresponding luminance sample positions in the preceding frame. The
sample values in these sample positions are made available by delaying the
preceding frame by one frame period, and hence the sample value in the
central position is referenced E0D, and the other eight sample values will
be referred to as S1D to S8D, respectively.
FIG. 3, to which reference is now made, shows part of the temporal error
concealment arrangement 5 of FIG. 1 in more detail. In order to make the
primary decision whether to use temporal or spatial concealment, the
sample values S1 to S8 and E0 in the nine sample positions of FIG. 2,
together with the sample values S1D to S8D and E0D in the corresponding
sample positions in the preceding frame, have to be made available sample
position by sample position for each data word received from the error
correction arrangement 2 of FIG. 1.
This is done by a frame store 10 which causes a delay of one frame period,
four line delays 11 to 14 each of which causes a delay of one horizontal
scan line period, and twelve sample delays 15 to 26, each of which causes
a delay of one sample period. Using the notation of FIG. 2, the delay
elements 10 to 26 together make available the eighteen sample values E0,
E0D, S1 to S2 and S1D of FIG. 2 at the points in the circuit indicated in
FIG. 3.
In order to decide whether temporal replacement is appropriate, the sample
values S1 to S8 in each of the eight sample positions nearest to the
current sample position are compared in value with the corresponding
sample values S1D to S8D from the preceding frame to obtain a measure of
the difference in their levels. This in effect a way of obtaining a
measure of the degree of movement in the picture from frame to frame. To
obtain this measure, the sample values S1 to S8 and S1D to S8D are
supplied in corresponding pairs to eight subtractors 27 to 34. Thus, the
subtractor 27, for example, receives the sample values S1 and S1D and
supplies an output corresponding to their difference in level to a
threshold programmable read-only memory (PROM) 35. Likewise the
subtractors 28 to 34 supply outputs corresponding to the differences in
the levels of the sample values respectively applied to them, to
respective threshold PROMs 36 to 42.
Each of the threshold PROMs 35 to 42 operates to supply a scaled output
representing the signal supplied thereto by the corresponding subtractor
27 to 34, and as the signal supplied to each of the threshold PROMs 35 to
42 increases, the quantization becomes coarser. In other words, this
scaling is effected so that outputs from the subtractors 27 to 34
representing small differences in the two sample values supplied to the
respective subtractors 27 to 34 result in small scaled outputs from the
threshold PROMs 36 to 42, but as the difference in the two sample values
increase, the scaled outputs from the threshold PROMs 36 to 42 increase
more slowly.
It will be recalled that each of the input data words which are supplied by
way of an input 43 comprises eight data bits and a ninth bit which forms
an error flag and which has been added by the error correction arrangement
2 (FIG. 1). Considering only the threshold PROM 35 for the moment, the
error flags associated with the data words representing the sample values
S1 and S1D supplied to the subtractor 37 are derived by an OR-gate 44. If
either or both of the error flags is set to indicate that the
corresponding sample value S1 or S1D is in error, the OR-gate 44 supplies
an error output to the threshold PROM 35 and also to one of eight inputs
of an error analysis PROM 45. If the threshold PROM 35 receives such an
error output from the OR-gate 45, the output of the threshold PROM 35 is
forced to zero, independent of the difference value supplied to the
threshold PROM 35 by the subtractor 27. Although for simplicity they are
not shown in FIG. 3, each of the threshold PROMs 36 to 42 likewise has an
associated OR-gate corresponding to the OR-gate 45 and operating in like
manner in dependence on the error flags associated with the sample values
supplied to the respective subtractors 28 to 34. Thus, if any of the
sample values S1 to S8 and S1D to S8D are in error, the output of the
appropriate threshold PROM 35 to 42 is forced to zero and moreover an
error output is supplied to the corresponding input of the error analysis
PROM 45.
The outputs of each of the threshold PROMs 35 to 42 are connected to an
adder 46 which adds the values of the eight inputs received and supplies
an output to a scaling PROM 47, which also receives an input from the
error analysis PROM 45. The scaling PROM 47 operates to give an overall
measure of the advisability of using temporal replacement for concealment.
Thus, if for example there were no movement in the picture and no errors
in any of the samples S1 to S8 and S1D to S8D, the output of the adder 46
would correspond to zero and there would be no error output from the error
analysis PROM 45. However, if there were movement in the picture, then the
output of the adder 46 would correspond to some value greater than zero.
Likewise, where errors are present, so that the error analysis PROM 45 is
supplying an output to the scaling PROM 47, the operation of the scaling
PROM 47 is such as effectively to increase any value which is being
received from the adder 46. This operation is such that the scaling PROM
47 supplies its maximum output when the error analysis PROM 45 is
receiving eight input error signals, even although in that case the output
of the adder 46 is zero, because the output of each of the threshold PROMs
35 to 42 has been forced to zero.
The output of the scaling PROM 47 is supplied to a further threshold PROM
48. The further threshold PROM 48 also receives as inputs the error flag
corresponding to the sample value E0 at the current sample position and
the error flag corresponding to the delayed sample value E0D from the
corresponding sample position in the preceding frame. The decision whether
to substitute the delayed sample value E0D for the sample value E0 is made
by the further threshold PROM 48. In other words, the threshold PROM 48
makes the primary decision whether to effect temporal replacement for
concealment at the current sample position or not. This decision is based
on three factors. Firstly, the error flag for the sample value E0 must
indicate that it is in error, because if it does not then no concealment,
whether temporal or spatial, is necessary. Secondly, the error flag
corresponding to the delayed sample value E0D must indicate that it is
error free, because if it is in error then it is not desired to use the
erroneous delayed sample value E0D as a replacement. Thirdly, the output
value supplied by the scaling PROM 47 must be below a predetermined
threshold value which is chosen to correspond to a reasonably low level of
movement and of sample values in error in that part of the television
picture corresponding to the eight sample positions surrounding the
current sample position. Where all three of these conditions are
satisfied, the further threshold PROM 48 supplies an output to a 2:1
selector 49 to indicate that replacement is to take place, whereupon the
selector 49 switches to supply the delayed sample value E0D to an output
50. At all other times the sample value E0 is supplied by the selector 49
to the output 50.
Referring again to FIG. 1, the overall operation is therefore such that if
the current sample value is not flagged as being in error, it is passed to
the output terminal 4. If the current sample value is flagged as being in
error, and if the primary decision of the temporal error concealment
arrangement 5 is that temporal replacement should take place, then in
place of the current sample the correspondingly positioned sample value
E0D delayed by one frame is supplied to the output terminal 4. If,
however, the temporal error concealment arrangement 5 decides that
although the current sample value E0 is in error, temporal replacement is
not appropriate, because of the number of errors in the immediately
adjacent sample values and/or because movement in the picture is causing
substantial changes in the sample values in corresponding positions from
frame to frame, then the decision passes to the spatial error concealment
arrangement 6.
The spatial error concealment arrangement 6 can take any one of a number of
different known forms, for example that disclosed in our UK Pat. No. 2 084
432 referred to above. However, it is preferred that the spatial error
concealment arrangement 6 combines several known methods of spatial
concealment, and works through these methods in sequence to select the one
most appropriate to the given current picture content and error situation.
The preferred sequence is as follows:
1. Spatial concealment, the concealment being adaptive by picture analysis.
2. Spatial concealment, the concealment being adaptive by error analysis.
3. Spatial replacement, the replacement being adaptively steered using
spatial and error analysis.
All these methods are known, but an example of each method will be briefly
described. Firstly, spatial concealment where the concealment is adaptive
by picture analysis may comprise generating a concealment sample value by
averaging the two or more sample values most closely adjacent to the
current sample value in any one of the available four directions, that is
the horizontal, the vertical and the positive and negative diagonal
directions, in the same field or frame. This averaging in the four
directions is done continuously to maintain a current indication of which
direction of averaging is giving the best results, and when the current
sample value is flagged as being in error, and the primary decision is not
to effect temporal replacement, spatial concealment is effected using the
direction currently indicated as giving the best result.
Secondly, spatial concealment where the concealment is adaptive by error
analysis may comprise generating a concealment sample value by averaging
the two or more sample values most closely adjacent to the current sample
value in any one of the available four directions, that is the horizontal,
the vertical and the positive and negative diagonal directions, in the
same field or frame. This averaging in the four directions is done
continuously to maintain a current indication of which direction of
averaging involves the use of the fewest sample values flagged as being in
error. When two directions both involve the use of the lowest number of
sample values which are in error, then the direction using the spatially
nearest sample values, which will generally be the horizontal direction,
is indicated as giving the best result. Then, when the current sample
value is flagged as being in error, and the primary decision is not to
effect temporal replacement, spatial concealment is effected using the
direction currently indicated as giving the best result.
Thirdly, spatial replacement where the replacement is adaptively steered
using spatial and error analysis may comprise generating a replacement
value simply by taking the sample value most closely adjacent to the
current sample value in any one of the available eight directions in the
same field or frame. This replacement from the eight directions is tested
continously to maintain a current indication of which direction is giving
the best result, both in terms of not using sample values which are
flagged as being in error and in terms of resulting in the selection of a
replacement sample value closest in level to the current sample value.
Then, when the current sample value is flagged as being in error, the
primary decision is not to effect temporal replacement, and error
concealment by adaptive steering is not possible, spatial replacement is
effected using the direction currently indicated as giving the best
result.
The spatial error concealment arrangement 6 sets a threshold level for each
of these methods of spatial replacement, and if none of the methods taken
in sequence reaches the predetermined threshold level and hence is likely
to result in satisfactory spatial concealment, then the spatial error
concealment arrangement 6 supplies an output to the temporal error
concealment arrangement 5, and the temporal error concealment arrangement
5 is operative to replace the error sample value E0 with the delayed
sample value E0D. It is recognized that this will give rise to a
temporally incorrect sample value, and hence may well result in a trailing
dot in the reproduced picture, but this is probably more acceptable than
leaving the error sample value E0 or replacing it with some other
erroneous sample value derived spatially.
The foregoing sequence of steps is used to decide upon the form of
replacement to be used when the DVTR 1 is operating in the normal playback
mode and in still and slow motion modes. However, when the DVTR 1 is
operating in the shuttle mode, the reproduced data is so imperfect and
relates to parts of so many frames compressed in time, that the primary
step of temporal concealment is omitted. In the shuttle mode, therefore,
the sequence becomes spatial concealment, the concealment being adaptive
by picture analysis; spatial concealment, the concealment being adaptive
by error analysis; spatial replacement, the replacement being adaptively
steered using spatial and error analysis; and, finally, temporal
replacement.
In the foregoing description it has been assumed that the selection of the
method of replacement to be used is done using the luminance sample
values. It is additionally possible to effect the same method for the
chrominance sample values in a composite digital television signal, but as
it is usually the case that a change in the luminance sample values is
accompanied by a corresponding change in the chrominance sample values, it
is simpler to effect concealment of the chrominance sample values using
the method selected in dependence on the examination of the luminance
sample values.
Various modifications are of course possible, and in particular the
operations performed by the elements 27 to 48 of FIG. 3 can be embodied in
different forms without departing from the scope of the invention as
defined in the broadest of the appended claims.
* * * * *
|
|
 |
|
 |
Description |
|
