 |
Claims  |
|
|
I claim:
1. A method of concealing errors in a digital television signal, which
television signal represents a raster comprising a plurality of component
sample signals corresponding respectively to sample positions along
horizontal scan lines of a television picture made up of a plurality of
such lines, the method comprising, in respect of each said sample signal
which is in error:
selecting from a plurality of algorithms a preferred algorithm for
correcting said error sample signal;
calculating a correction value for said error sample signal using said
preferred algorithm;
substituting said correction value for said error sample signal so as to
conceal the error; and
when the density of other error sample signals in the neighborhood of said
error sample signal prevents said selection of said preferred algorithm;
substituting the sample signal from a position in said raster adjacent to
the position of said error sample signal for said error sample signal so
as to conceal the error.
2. A method according to claim 1 wherein said adjacent position is selected
in dependence on the positions of other error sample signals in the
neighborhood of said error sample signal.
3. A method according to claim 2 wherein said adjacent position is in the
direction in said raster from the position of said error sample signal
determined by evaluation of directional accuracy factors E.sub.H, E.sub.V,
E.sub.D+ and E.sub.D- corresponding to the horizontal, vertical, positive
diagonal and negative diagonal directions respectively of said raster,
said factors being given by:
E.sub.H =.vertline.K.sub.H [S.sub.0,-2 -S.sub.0,-1 ]/K.sub.h .vertline.
E.sub.V =.vertline.K.sub.V [S.sub.-2,0 -S.sub.-1,0 ]/K.sub.v .vertline.
E.sub.D+ =.vertline.K.sub.D [S.sub.-2,2 -S.sub.-1,1 ]/K.sub.d+ .vertline.
E.sub.D- =.vertline.K.sub.D [S.sub.-2,-2 -S.sub.-1,-1 ]/K.sub.d- .vertline.
where:
K.sub.H, K.sub.V and K.sub.D are respective spatial weighting coefficients
which are proportional to the distances between the two sample positions
used in the respective expression;
S.sub.m,n is the value of the sample signal in lime m at position n
relative to said error sample signal S.sub.0,0 ; and
K.sub.h, K.sub.v, K.sub.d+ and K.sub.d- are error weighting coefficients
for the horizontal, vertical, positive diagonal and negative diagonal
directions respectively and which are determined in dependence on whether
none, one or two of the sample signals used in the respective expression
are in error.
4. A method according to claim 2 wherein the number of successive said
substitutions from an adjacent position of the same sample signal is
limited to a predetermined maximum.
5. A method according to claim 2 wherein said sample signals are stored in
a field store until over-written by a later said sample signal, the number
of successive said substitutions from an adjacent position of the same
sample signal is limited to a predetermined maximum, and on occurrence of
said limitation, the sample signal already stored at the address in said
field store corresponding to the position of said error sample signal
remains without being over-written.
6. A method according to claim 1 wherein said selection of a preferred
algorithm comprises:
using a first algorithm to calculate from available sample signals the
expected value of a first sample signal corresponding to a first sample
position adjacent to the sample position of said error sample signal;
checking said expected value of said first sample signal against the actual
value of said first sample signal;
using a second algorithm to calculate from available sample signals the
expected value of a second sample signal corresponding to a second sample
position adjacent to the sample position of said error sample signal;
checking said expected value of said second sample signal against the
actual value of said second sample signal; and
selecting one of said first and second algorithms in dependence on the
results of said checking steps and using said selected algorithm to
calculate from available sample signals said corrected value of said error
sample signal;
said first and second algorithms using available sample signals located
along respective different directions of said raster.
7. Apparatus for concealing errors in a digital television signal, which
television signal represents a raster comprising a plurality of component
sample signals corresponding respectively to sample positions along
horizontal scan lines of a television picture made up of a plurality of
such lines, the apparatus comprising:
means operative in respect of each said sample signal which is in error to
select from a plurality of algorithms a preferred algorithm for correcting
said error sample signal;
means for calculating a correction value for said error sample signal using
said preferred algorithm;
means for substituting said correction value for said error sample signal
so as to conceal the error; and
means operative when the density of other error sample signals in the
neighborhood of said error sample signal prevents said selection of said
preferred algorithm, to substitute the sample signal from a position in
said raster adjacent to the position of said error sample signal for said
error sample signal so as to conceal the error.
8. Apparatus according to claim 7 wherein said adjacent position is
selected in dependence on the positions of other error sample signals in
the neighborhood of said error sample signal.
9. Apparatus according to claim 8 wherein said adjacent position is in the
direction in said raster from the position of said error sample signal
determined by evaluation of directional accuracy factors E.sub.H, E.sub.V,
E.sub.D+ and E.sub.D- corresponding to the horizontal, vertical, positive
diagonal and negative diagonal directions respectively of said raster,
said factors being given by:
E.sub.H =.vertline.K.sub.H [S.sub.0,-2 -S.sub.0,-1 ]/K.sub.h .vertline.
E.sub.V =.vertline.K.sub.V [S.sub.-2,0 -S.sub.-1,0 ]/K.sub.v .vertline.
E.sub.D+ .vertline.K.sub.D [S.sub.-2,2 -S.sub.-1,1 ]/K.sub.d+ .vertline.
E.sub.D- =.vertline.K.sub.D [S.sub.-2,-2 -S.sub.-1,-1 ]/K.sub.d- .vertline.
where:
K.sub.H, K.sub.V and K.sub.D are respective spatial weighting coefficients
which are proportional to the distances between the two sample positions
used in the respective expression;
S.sub.m,n is the value of the sample signal in line m at position n
relative to said error sample signal S.sub.0,0 ; and
K.sub.h, K.sub.v, K.sub.d+ and K.sub.d- are error weighting coefficients
for the horizontal, vertical, positive diagonal and negative diagonal
directions respectively and which are determined in dependence on whether
none, one or two of the sample signals used in the respective expression
are in error.
10. Apparatus according to claim 8 comprising counter means for limiting
the number of successive said substitutions from an adjacent position of
the same sample signal to a predetermined maximum.
11. Apparatus according to claim 8 comprising a field store in which said
sample signals are stored until over-written by a later said sample
signal, and counter means for limiting the number of successive said
substitutions from an adjacent position of the same sample signal to a
predetermined maximum, on occurrence of said limitation, the sample signal
already stored at the address in said field store corresponding to the
position of said error sample signal remaining without being over-written.
12. Apparatus according to claim 7 wherein said means operative to select a
preferred algorithm comprises:
means operative in respect of each said sample signal which is in error to
use a first algorithm to calculate from available sample signals the
expected value of a first sample signal corresponding to a first sample
position adjacent to the sample position of said error sample signal;
means to check said expected value of said first sample signal against the
actual value of said first sample signal;
means to use a second algorithm to calculate from available sample signals
the expected value of a second sample signal corresponding to a second
sample position adjacent to the sample position of said error sample
signal;
means to check said expected value of said second sample signal against the
actual value of said second sample signal; and
means to select one of said first and second algorithms in dependence on
the results of said checking steps and using said selected algorithm to
calculate from available sample signals said corrected value of said error
sample signal;
said first and second algorithms using available sample signals located
along respective different directions of said raster.
13. A method of concealing errors in a digital television signal, which
television signal represents a raster comprising a plurality of component
sample signals corresponding to respective sample positions along
horizontal scan lines of a television picture made up of a plurality of
such lines, the method comprising, in respect of each said sample signal
which is in error, the steps of:
generating a plurality of algorithms for potentially correcting said error
sample signal;
selecting one of said algorithms as better suited than the others for
correcting said error sample signal;
determining the density of other error sample signals adjacent said error
sample signal; and
substituting for said error sample signal, in response to the determination
of the density of said other error sample signals, either a correction
sample signal related to said selected algorithm or a sample signal from a
position in said raster adjacent to the position of said error sample
signal, whereby to correct said error.
14. The method of claim 13; wherein said step of substituting in response
to said determination includes the step of selecting said adjacent
position in dependence on the positions of other error sample signals
adjacent said error sample signal.
15. The method of claim 14; wherein said step of substituting in response
to said determination includes the step of substituting the same signal
from an adjacent position fewer than a predetermined number of times.
16. The method of claim 14; and further comprising the steps of:
storing said sample signals in field store means;
counting the number of successive substitutions from an adjacent position
of the same sample signal; and
selectively overwriting said sample signals stored in said field store
means with a later sample signal, said sample signal already stored in
said field store means corresponding to the portion of said error sample
signal remaining without being overwritten when said number of successive
substitutions exceeds a predetermined number.
17. The method of claim 14; wherein said step of selecting said adjacent
position includes the step of determining the direction in said raster
from the position of said error sample signal by calculating directional
accuracy factors E.sub.H,E.sub.V,E.sub.D+ and E.sub.D- corresponding to
the horizontal, vertical, positive diagonal and negative diagonal
directions, respectively, of said raster, said factors being given by:
E.sub.H =.vertline.K.sub.H [S.sub.0,-2 -S.sub.0,-1 ]/K.sub.h .vertline.
E.sub.V =.vertline.K.sub.V [S.sub.-2,0 -S.sub.-1,0 ]/K.sub.v .vertline.
E.sub.D+ =.vertline.K.sub.D [S.sub.-2,2 -S.sub.-1,1 ]/K.sub.d+ .vertline.
E.sub.D =.vertline.K.sub.D [S.sub.-2,-2 -S.sub.-1,-1 ]/K.sub.d- .vertline.
where:
K.sub.H, K.sub.V and K.sub.D are respective spatial weighting coefficients
which are proportional to the distances between the two sample positions
used in the respective expression;
S.sub.m,n is the value of the sample signal in line m at position n
relative to said error sample signal S.sub.0,0 ; and
K.sub.h, K.sub.v, K.sub.d+ and K.sub.d- are error weighting coefficients
for the horizontal, vertical, positive diagonal and negative diagonal
directions, respectively, and which are determined in dependence on the
number of the sample signals used in the respective expression which are
in error.
18. The method of claim 13; wherein said step of selecting one of said
correcting error sample signals comprises the steps of:
generating from available sample signals an expected value signal for a
first sample signal corresponding to a first sample position adjacent to
the sample position of said error sample signal;
comparing said expected value signal for said first sample signal against
the actual value signal for said first sample signal to generate a first
comparison signal;
generating from available sample signals an expected value signal for a
second sample signal corresponding to a second sample position adjacent to
the sample position of said error sample signal;
comparing said expected value signal for said second sample signal against
the actual value signal for said second sample signal to generate a second
comparison signal; and
selecting one of said expected value signals for said first and second
sample signals in response to said first and second comparison signals to
generate from available sample signals said corrected value of said error
sample signal;
said expected value signals for said first and second sample signals using
available sample signals located along respective different directions of
said raster.
19. Apparatus for concealing errors in a digital television signal, which
television signal represents a raster comprising a plurality of component
sample signals corresponding to respective sample positions along
horizontal scan lines of a television picture made up of a plurality of
such lines, the apparatus comprising, in respect of each said sample
signal which is in error:
means for generating a plurality of algorithms for potentially correcting
said error sample signal;
means for selecting one of said algorithms as better suited than the others
for correcting said error sample signal; and
means for determining the density of other error sample signals near said
error sample signal and for substituting, in response to such
determination, either a correction sample signal related to said selected
algorithm or a sample signal from a position in said raster adjacent to
the position of said error sample signal for said error sample signal so
as to conceal the error.
20. The apparatus of claim 19; wherein said means for determining includes
means for selecting said adjacent position in dependence on the positions
of other error sample signals adjacent said error sample signal.
21. The apparatus of claim 20; wherein said means for selecting includes
means for calculating said adjacent position in the direction in said
raster from the position of said error sample signal by calculating
directional accuracy factors E.sub.H, E.sub.V, E.sub.D+ and E.sub.D-
corresponding to the horizontal, vertical, positive diagonal and negative
diagonal directions, respectively, of said raster, said factors being
given by:
E.sub.H =.vertline.K.sub.H [S.sub.0,-2 -S.sub.0,-1 ]/K.sub.h .vertline.
E.sub.V =.vertline.K.sub.V [S.sub.-2,0 -S.sub.-1,0 ]/K.sub.v .vertline.
E.sub.D+ =.vertline.K.sub.D [S.sub.-2,2 -S.sub.-1,1 ]/K.sub.d+ .vertline.
E.sub.D- =.vertline.K.sub.D [S.sub.-2,-2 -S.sub.-1,-1 ]/K.sub.d- .vertline.
where:
K.sub.H, K.sub.V and K.sub.D are respective spatial weighting coefficients
which are proportional to the distances between the two sample positions
used in the respective expression;
S.sub.m,n is the value of the sample signal in line m at position n
relative to said error sample signal S.sub.0,0 ; and
K.sub.h, K.sub.v, K.sub.d+ and K.sub.d- are error weighting coefficients
for the horizontal, vertical, positive diagonal and negative diagonal
directions, respectively, and which are determined in dependence on the
number of the sample signals used in the respective expression which are
in error.
22. The apparatus of claim 20; further comprising counter means for
limiting the number of successive said substitutions from an adjacent
position of the same sample signal to a predetermined maximum.
23. The apparatus of claim 20; further comprising field store means for
storing said sample signals until over-written by a later said sample
signal, and counter means for limiting the number of successive said
substitutions from an adjacent position of the same sample signal to a
predetermined maximum, and on occurrence of said limitation, the sample
signal already stored at the address in said field store means
corresponding to the position of said error sample signal remaining
without being over-written.
24. The apparatus of claim 19; wherein said means for selecting one of said
algorithms comprises:
means operative in respect of each said sample signal which is in error for
calculating from available sample signals an expected value signal for a
first sample signal corresponding to a first sample position adjacent to
the sample position of said error sample signal;
means for comparing said expected value signal for said first sample signal
with an actual value signal for said first sample signal to generate a
first comparison signal;
means for calculating from available sample signals an expected value
signal of a second sample signal corresponding to a second sample position
adjacent to the sample position of said error sample signal;
means for comparing said expected value signal for said second sample
signal with an actual value signal for said second sample signal to
generate a second comparison signal; and
means for selecting one of said expected value signals for said first and
second sample signals in dependence on the results of said first and
second comparison signals to generate from available sample signals said
corrected value of said error sample signal;
said expected value signals for said first and second sample signals using
available sample signals located along respective different directions of
said raster. |
|
 |
|
 |
Claims |
|
|
|
Description |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to error concealment in digital television signals.
2. Description of the Prior Art
Recently there has been an increasing interest in the use of digital
techniques for television signals. Such techniques are, for example, used
in some video tape recording arrangements where an incoming television
signal to be recorded is sampled, the samples are coded into digital form,
the digital data signals are recorded and subsequently reproduced by a
video tape recorder (VTR), 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 drop-out occurring in the VTR, the digital signals are lost or
corrupted and then the reformed television signal does not correspond
exactly to the original television signal, and a resulting television
picture is degraded.
There are two main approaches to dealing with errors in digital television
signals. The first approach is correction, which 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
correction capability would require an excessive amount of additional data
which might overload the data handling paths or raise the data rate to an
unacceptable level. The second approach, with which the present invention
is more particularly concerned, is concealment. This comprises the
replacement of corrupted data signals by data signals generated using
available uncorrupted data signals. This method relies largely for
accuracy on the strong correlation that exists in a television signal.
In our co-pending UK patent application No. 8011090 (Ser. No. 2073534), the
corresponding co-pending European patent application No. 81301156.6 (Ser.
No. 0037212), and corresponding U.S. Pat. No. 4,419,693 we have disclosed
a method of error concealment which comprises selecting from a plurality
of algorithms a preferred algorithm for calculating a corrected value for
use in concealment of an error sample, calculating a corrected value for
the sample using the preferred algorithm, and replacing the error sample
by the corrected value sample. This method works well at the normal
reproduction speed, but when reproducing at higher speeds the loss and
corruption of reproduced signals as a reproducing head crosses recorded
tracks means that not only is correction substantially less effective, but
so also is concealment. This is because concealment relies upon the
presence of valid samples adjacent to a sample to be concealed, and if a
significant number of the adjacent samples have been lost or corrupted,
effective concealment is impossible. This problem will be described
further below.
SUMMARY OF THE INVENTION
One object of the present invention is to provide a method of concealing
errors in a digital television signal and which overcomes this problem.
Another object of the present invention is to provide a method of
concealing errors in a digital television signal which is operable in the
presence of a high density of error sample signals.
Another object of the present invention is to provide a method of and
apparatus for concealing errors in a digital television signal by
substitution of actual sample signals when the density of errors is too
high to permit calculation of a sample value for substitution.
According to the present invention there is provided a method of concealing
errors in a digital television signal, which television signal represents
a raster comprising a plurality of component sample signals corresponding
respectively to sample positions along horizontal scan lines of a
television picture made up of a plurality of such lines, the method
comprising, in respect of each sample signal which is in error:
selecting from a plurality of algorithms a preferred algorithm for
correcting error sample signal;
calculating a corrected value of the error sample signal using the
preferred algorithm;
substituting the corrected sample signal for the error sample signal so as
to conceal the error; and
when the density of other error sample signals in the neighborhood of the
error sample signal prevents the selection of the preferred algorithm;
substituting the sample signal from a position in the raster adjacent to
the position of the error sample signal for the error sample signal so as
to conceal the error.
According to the present invention there is also provided apparatus for
concealing errors in a digital television signal, which television signal
represents a raster comprising a plurality of component sample signals
corresponding respectively to sample positions along horizontal scan lines
of a television picture made up of a plurality of such lines, the
apparatus comprising:
means operative in respect of each sample signal which is in error to
select from a plurality of algorithms a preferred algorithm for correcting
the error sample signal;
means for calculating a corrected value of the error sample signal using
the preferred algorithm;
means for substituting the corrected sample signal for the error sample
signal so as to conceal the error; and
means operative when the density of other error sample signals in the
neighbourhood of the error sample signal prevents the selection of the
preferred algorithm, to substitute the sample signal from a position in
the raster adjacent to the position of the error sample signal for the
error sample signal so as to conceal the error.
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 shows a matrix of sample positions in a television picture;
FIG. 2 shows in simplified block form apparatus for concealing errors in a
digital television signal;
FIG. 3 shows another matrix of sample positions in a television picture;
FIG. 4 shows a decision tree for a method according to the invention;
FIG. 5 is a diagram of a matrix of samples in a television picture; and
FIG. 6 is a more detailed block diagram of a portion of FIG. 2.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
Before describing an embodiment of the invention, and to assist
understanding of the embodiment, further reference will first be made to
the problem mentioned above.
Referring to FIG. 1, this shows part of a television raster, and in
particular parts of three consecutive horizontal scan lines labelled line
n-1, line n and line n+1. The sample positions are disposed at regular
intervals along each of the lines, the intervals corresponding to a
sampling frequency of say 13.5 MHz, and the sample positions being aligned
in the vertical direction. Reading from the left, consecutive sample
positions in each line are labelled S-3, S-2, S-1, S0, S1, S2 and S3.
Using this notation, any sample position in the matrix can be designated
by the line and the sample number, and for the purpose of this discussion
it is assumed that the sample position at which there is an error sample
signal requiring concealment is in line n at position S0, this being
designated n, S0.
As disclosed in our above-mentioned applications, a corrected value for the
sample position n, S0 could be estimated in one of four different ways.
Firstly, the average could be taken of the two samples in line n adjacent
to and on each side of the sample position n, S0. Secondly, the average
could be taken of the two sample values in line n-1 and line n+1 adjacent
to and vertically above and below the sample position n, S0. Thirdly, the
average could be taken of the two sample values in line n-1 and line n+1
and on either side of the sample position n, S0 along the positive
diagonal direction. Fourthly, the average could be taken of the two sample
values in line n-1 and line n+1 adjacent to and on either side of the
sample position n, S0 and along the negative diagonal direction. These
four directions are indicated by the arrows A, B, C and D respectively.
Each of these possibilities may be thought of as an algorithm for
calculating a corrected value, and it will be appreciated that it is
likely that one of these algorithms will give a better result than any of
the others. The direction to be used is therefore selected by testing each
algorithm using known sample values to see which gives the best result,
and then using a corrected value derived using the direction corresponding
to that preferred algorithm when substituting a corrected value sample.
As a further refinement, the results derived from the respective algorithms
can be weighted. In other words, a value can be placed on the likely
accuracy of the results obtained. This is necessary because the distance
between adjacent sample positions is less in the horizontal direction than
in the vertical direction, the difference amounting to a factor of
approximately 1.8. For this reason, the algorithm using the horizontal
direction is in fact most likely to give the nearest result, with the
algorithm for the vertical direction being next best, and the two
algorithms for the diagonal directions being the next best.
The four algorithms referred to above will now be specified in mathematical
terms. Thus, the decision of concealment direction is made by
investigating the adjacent sample values and obtaining the concealment
accuracy for each direction. If the concealment accuracy is H for the
horizontal direction, V for the vertical direction, D.sup.+ for the
positive diagonal direction and D.sup.- for the negative diagonal
direction, then these concealment accuracies can be defined as follows:
H=1/2.vertline.1/2[(n-1),S-1+(n-1),S1]-(n-1),S0.vertline.+1/2.vertline.1/2[
(n+1),S-1+(n+1),S1]-(n+1),S0.vertline. (1)
that is to say, the concealment accuracy H equals the average of the
horizontal concealment accuracy from the horizontal line immediately above
and the horizontal line immediately below the horizontal line containing
the error sample. Likewise:
V=1/2.vertline.1/2[(n-1),S-1+(n+1),S-1]-n,S-1.vertline.+1/2.vertline.1/2[(n
+1),S1+(n+1),S1]-n,S1.vertline. (2)
D.sup.+
=1/2.vertline.1/2[(n-1),S0+(n+1),S-2]-n,S-1.vertline.+1/2.vertline.1/2[(n-
1),S2+(n+1),S0]-n,S1.vertline. (3)
D.sup.-
=1/2.vertline.1/2[(n-1),S-2+(n+1),S0]-n,S-1.vertline.+1/2.vertline.1/2[(n-
1),S0+(n+1),S2]-n,S1.vertline. (4)
These four values H, V, D.sup.+ and D.sup.- represent the accuracy of
concealment for the sample values most closely connected with the error
sample. Preferably these values are each assigned a weighting coefficient
to take account of the unequal spacings of the horizontal, vertical and
diagonal samples. The smallest value is then used to select the direction
of concealment.
The method has been described as applied to the luminance channel, that is
to say concealment of errors occurring in luminance sample values. It is
also necessary to consider the colour difference channels, and here two
possibilities arise.
Firstly, each color difference channel can be provided with a separate
concealment selection arrangement independent of the arrangement for the
luminance channel.
Secondly, because the first solution referred to above increases the amount
of hardware required by approximately three, an alternative method which
economizes on the amount of hardware required makes use of the fact that
the chrominance information is related to the luminance information. That
is, where a chrominance edge exists, so usually does a luminance edge.
Based on this assumption it is possible to select the direction of color
difference concealment to be the same as that selected for luminance
concealment. However, because the chrominance samples occur at only one
half the frequency of the luminance samples along each horizontal line, a
different set of weighting coefficients has to be used, these being
optimized to the chrominance bandwidths.
Referring to FIG. 2, this shows apparatus for concealing errors in a
digital television signal. The apparatus comprises a luminance sample
storage means 1 to which luminance input samples are supplied by way of an
input terminal 2. The luminance sample storage means 1 supplies outputs to
a luminance sample matrix storage means 3 which stores a moving matrix of
sample values corresponding to the sample positions (n+1),S2; (n+1),S1;
(n+1),S0; (n+1),S-1; (n+1),S-2; n,S1; n,S0; n,S-1; (n-1),S2; (n-1),S1;
(n-1),S0; (n-1),S-1; and (n-1),S-2.
Four concealment accuracy detectors are provided, these being a horizontal
concealment accuracy detector 4, a vertical concealment accuracy detector
5, a positive diagonal concealment accuracy detector 6 and a negative
diagonal concealment accuracy detector 7. Each of the concealment accuracy
detectors 4 to 7 is continuously supplied with the appropriate part of the
sample matrix from the luminance sample matrix storage means 3. Thus the
horizontal concealment accuracy detector 4, for example, receives or
selects the sample values necessary to calculate the concealment accuracy
H using algorithm (1) above, and supplies a signal representing the
concealment accuracy H by way of a weighting multiplier 8 to a luminance
direction processor 12. Likewise the concealment accuracy detectors 5 to 7
supply a respective signal representing the vertical concealment accuracy
V, the positive diagonal concealment accuracy D.sup.+ and the negative
diagonal concealment accuracy D.sup.- by way of weighting multipliers 9,
10 and 11 respectively to the luminance direction processor 12. The
weighting multipliers 8 to 11 effect the weighting referred to above to
compensate for the different distances between adjacent sample positions
in the various directions. The weighting may be done simply on the basis
of distance between adjacent sample positions, in which case each
weighting multiplier multiplies by the reciprocal of the distance between
adjacent sample positions in the relevant direction. Other weightings can,
however, be used.
The luminance direction processor 12 supplies an output signal representing
the selected direction of concealment to a sample value calculator 13
which operates to select the appropriate samples from the luminance sample
matrix storage means 3 and calculate therefrom the required concealment
value to be used to conceal the error sample. For example, if the
horizontal direction is selected, the sample value calculator 13 uses the
sample values for the sample positions n,S-1 and n,S1 to calculate the
value to be used to conceal the error sample at the sample position n, S0.
The concealment value is supplied to a selector 14 to which a switching
signal is supplied by way of a terminal 15. The selector 14 is also
supplied with the sample value from the sample position n,S0 by way of a
terminal 16.
Preferably the apparatus as so far described operates continuously, that is
to say concealment values are determined as described for every sample
position and supplied to the selector 14. Only, however, when it has been
determined that there is an error at a given sample position n,S0, is a
signal supplied to the selector 14 by way of the terminal 15, whereupon
the concealment value supplied from the calculator 13 is supplied to a
luminance output terminal 17 in place of the sample value supplied by way
of the terminal 16. At all other times, the sample value supplied by way
of the terminal 17 is supplied to the luminance output terminal 17.
The fact that there is an error at a given sample position n,S0 can be
determined in any suitable manner. For example, it may be determined that
the data word representing the sample value is not valid. As a more
specific example, suppose that each sample value is coded into a word in
the sub-set of 10-bit words which consist of 5 "0" and 5 "1"; this being
convenient for magnetic recording and reproduction because of the large
number of transients and the ease of clock recovery. In this case any
reproduced data word not having 5 "0" and 5 "1" is not a valid member of
the sub-set and so is clearly an error. Thereupon a switching signal is
supplied to the terminal 15.
The apparatus may also include arrangements for calculating concealment
values for the color difference channels U and V. For simplicity, only
that part of the apparatus necessary to calculate concealment values for
the difference channel U is shown and will be described. For this purpose
the apparatus comprises a chrominance sample storage means 21 to which
chrominance input samples are supplied by way of an input terminal 22. The
chrominance sample storage means 21 supplies outputs to a chrominance
signal matrix storage means 23 which stores a moving matrix of sample
values corresponding to those listed above in connection with the
luminance sample matrix storage means 3, but adjusted to take account of
the different spacing between adjacent chrominance samples.
Operating in time division multiplex for the luminance and chrominance
samples respectively, the concealment accuracy detectors 4 to 7 derive
signals representing the horizontal, vertical, positive diagonal and
negative diagonal concealment accuracies H, V, D.sup.+ and D.sup.- for the
chrominance difference channel U and supply the signals by way of
respective chrominance weighting multipliers 24, 25, 26 and 27 to a
chrominance direction processor 28 which supplies an output signal
representing the selected direction of concealment to a sample value
calculator 29 which operates to select the appropriate samples from the
chrominance sample matrix storage means 23 and calculate therefrom the
required concealment value to be used to conceal the error sample. The
concealment error is supplied to a selector 30 to which a switching signal
is supplied by way of a terminal 31. The selector 30 is also supplied with
the sample value from the sample position n,S0 by way of a terminal 32.
As with the luminance part of the apparatus, the chrominance part of the
apparatus preferably operates continuously. Only, however, when it has
been determined there is an error at a given sample position n,S0, is a
signal supplied to the selector 30 by way of the terminal 31, whereupon
the concealment value supplied from the calculator 29 is supplied to a
chrominance output terminal 33 in place of the sample value supplied by
way of the terminal 32.
The chrominance part of apparatus may be duplicated for the color
difference channel V or alternatively hardware can be saved by also using
the algorithm selected for the color difference channel U for the color
difference channel V.
The method described above may be modified as described in our copending UK
patent application No. 8214086 (corresponding to U.S. application Ser. No.
494,324 filed May 13, 1983), to steer the algorithms used to avoid samples
known to be in error when calculating sample values for replacement.
Briefly, this is done by calculating each of the concealment accuracies
defined by expressions (1) to (4) in two component parts corresponding
respectively to the first and second lines of each of expressions (1) to
(4). Any calculations involving the use of an error sample is rejected so,
depending on the density of errors, a concealment accuracy may be
calculated using sample values on one side or the other side or both sides
of the sample to be concealed. With this modified method therefore the
algorithms are steered to avoid error samples using the fact that eight
different calculations for concealment accuracies are available.
Thus, in embodiments of the invention, in calculating a corrected value
sample no direction of concealment is used if the resulting calculation
involves the use of an error sample. In the present discussion it will be
assumed that all error samples are identified as such by having an error
flag, generally an additional bit "1", attached to the data word
representing that sample. However, it is not necessary to exclude a
direction of concealment merely because the concealment accuracy H, V,
D.sup.+ or D.sup.-+ as set out above is invalidated by one or more error
samples.
Consider, for example, the horizontal concealment accuracy H which is
calculated from the above algorithm (1) repeated here:
H=1/2.vertline.1/2[(n-1),S-1+(n-1),S1]-(n-1),S0.vertline.+1/2.vertline.1/2[
(n+1),S-1+(n+1),S1]-(n+1),S0.vertline. (1)
This algorithm can be viewed as the sum of two component algorithms
respectively bounded by the magnitude signs and it is possible for one of
these component algorithms to be invalidated by one or more error samples,
while the other component algorithm remains valid. Thus either of the
component algorithms may be dropped in favor of the other in appropriate
cases, and this results in two further algorithms for the horizontal
concealment accuracy, as follows:
H.sub.1 =.vertline.1/2[(n-1),S-1+(n-1),S1]-(n-1),S0.vertline.(5)
and
H.sub.2 =.vertline.1/2[(n+1),S-1+(n+1),S1]-(n+1),S0.vertline.(6)
For ease of subsequent consideration actual values will be ascribed to each
sample position as indicated in FIG. 5. Using these values the modified
horizontal concealment accuracies can be more simply expressed as:
H.sub.1 =.vertline.1/2(A1+A3)-A2.vertline. (7)
and
H.sub.2 =.vertline.1/2(A4+A6)-A5.vertline. (8)
This approach can be similarly applied to the other concealment accuracies
V, D.sup.+ and D.sup.-, for which modified concealment accuracies are
derived as follows:
V.sub.1 =.vertline.1/2[(n-1),S-1+(n | | |
