FR2513838A1 - Pre-emphasis sampling mode filter for SECAM TV colour coder - operates on chrominance subcarrier signal by digital sampling to achieve specified transfer function - Google Patents

Abstract

The filter has an 'inverted bell' response, intended for use with the chrominance signal of a SECAM colour television encoder. The filter operates by sampling, preferable digitally, and has a transfer function of the form (G(z)=a+bz power -1+cz power -2. The sampling frequency is four times the frequency of the chrominance sub-carrier in the PAL colour television system. The filter includes a first delay element (15) whose input is connected to the first inputs (20,18) of two adders (21,17) - one via a further delay line (19). The first adder output is fed directly to one input of a subtractor (28), while the second adder output passes via further processing circuitry to the second subtractor input. The subtractor output then forms the final output of the filter circuit.

Description

FILTER IN SAMPLE PREACCENTUATION MODE
OF A SUBCARRIER OF CHROMINANCE AND ENCODER
AND TELEVISION CAMERA COMPRISING SUCH A FILTER.

 The invention relates to a pre-emphasis filter of the chrominance signal for the Permission of television signals, in particular by the SECAM method, as well as to an encoder and a television camera comprising such a filter.

 In the SECAM process the chrominance signals are transmitted by frequency modulation, the carrier frequency, which is called the chrominance subcarrier, having the value f0 = 4,286 MHz. In such a transmission, the influence of the parasitic signals, ie the noise, is higher the greater the difference in frequency of the signal compared to that of the carrier, a protection against noise is provided by a filter, generally called "anticloche", whose amplitude / frequency response curve has a minimum for the frequency f0 = 4,286 MHz. Such a filter thus accentuates the signals furthest in frequency from the carrier, which makes it possible to standardize the signal / noise ratio.

 The anticlock filters known until now are of the analog type. But the need arises for filters of the type in digital sampled mode realized using integrated circuits in order firstly to allow the realization of a color television camera whose coder has a reduced weight and bulk and whose manufacturing cost is low and on the other hand to take advantage of the advantages inherent in digital coding, namely noise insensitivity and reproducibility.

 Knowing the transfer function of an analog filter, it is known, after calculating the z transfer function, to make the digital filter of the same transmittance using flip-flops, adders, subtracters and possibly gates.

 It has been found that the application as such to the anticlock filter of this method for synthesizing digital filters results in a result which is difficult to implement in practice owing to, in particular, its large number of elements and the large number of bits on which must be encoded the signal.

The filter according to the invention does not have these disadvantages. It is characterized in that its z-transfer function is of the form:
G (z) = a + bz'l + cz2 (1)
In this formula a, b and c are constants.

 A filter having this transfer function can be produced in a particularly simple manner and requires only a signal coded on "n limited number of bits, for example 8 bits.In addition, since the filter is not recursive, it introduces no instability. .

 As will be seen later, the invention is based on the observation that the transfer function of the corresponding analog filter can be simplified.

on conçoit que les coefficients peuvent être obtenus par la technique dite du décalage de bits qui est une simple modification de connexions n'introduisant aucun retard. In one embodiment, the z-transfer function of the digital anticlock filter is, by a factor:
H (z) = 1.25 - 0.125 -1 + 1.125 -2 (2)
Given that:

it is understood that the coefficients can be obtained by the so-called bit shift technique which is a simple modification of connections introducing no delay.

Other characteristics and advantages of the invention will become apparent with the description of some of its embodiments, this being done with reference to the attached drawings in which:
FIG. 1 is a diagram of filter response curves,
FIG. 2 is a filter diagram according to the invention,
FIGS. 3 and 4 are diagrams of parts of the filter of FIG. 2,
FIG. 5 is another diagram of filter response curves,
FIG. 6 is a diagram of a SECAM encoder of the digital type using the filter according to the invention, and
FIG. 7 is an anticlockwise digital filter diagram according to another embodiment.

The transfer function of an anticlock filter of the analog type

<tb> gique <SEP> is <SEP>: <SEP> w \
<tb><SEP> 1 <SEP> + <SEP> jQ <SEP>(<SEP> w <SEP> o
<tb><SEP> F (w) <SEP> = <SEP> w
<tb><SEP> 1 <SEP> + <SEP> jet <SEP> {ccr
<Tb>
In this formula:
w = 2 nf is the pulse, #
0 f0 = 2 # = 4,286 MHz (chrominance subcarrier
SECAM)
Q = 16 and Q '= 1.26.

To obtain the z-transfer function of a digital filter having the same response curve as this analog filter, in this formula the variable jw is replaced by the variable z-1.
z + 1 (this transformation is called a z-transformation of the homographic type) and we modify the constant terms to 0, Q and Q 'according to this transformation; we then obtain the following z-transfer function:
I 2
1 AZ-1 + BZ-2
1 - Cz + Dz with:
K = 8.3034; A = 0.09982; B = 0.90945; C = 0.06525; D = 0.24820 (6)
A digital filter having this z-transfer function will be described later in connection with FIG. 7. However, this filter is too complex for most applications, particularly for a public television television camera. to say that must be mass produced and cheap.

An anticlockwise digital filter of the type shown in FIG. 2, which will be described below, is the result of the following considerations:
It was first found, by calculation, that over the frequency range concerned: 3.9 to 4.75 MHz the influence of the factor Q 'on the quality coefficient of the filter whose transfer function is defined by the formula (4) above is negligible, that is to say, in this formula, we can keep only the numerator without the operation of the filter is affected.

 This observation is confirmed by FIG. 1, in which the frequency f in KHz has been plotted on the abscissa and the transmittance T in decibels (dB) of anticlockwise filters is ordinate. Z; a curve 10 in full line corresponds to the response of a filter whose transfer function is defined by formula (4) while the curve 11 in phantom corresponds to the response of a filter whose transfer function is the numerator of formula (4). It is clear that between 3.9 and 4.75 MHz the two curves 10 and 11 are practically merged.

To achieve a digital filter having the same response in the frequency range concerned, 3.9 to 4.75 MHz, we therefore start from the following transfer function:
w
N (p) = + p Q + Q p
p
In this formula p = jw
And we calculate the z-transform of the inverse of N (p), that is to say the transform of:

<tb> F (p) = <SEP> N1 <SEP> (p) <SEP> = <SEP>, <SEP> 1 <SEP> = <SEP> p
<tb><SEP> l + p <SEP> S <SEP> Qp0 <SEP> gQ <SEP> + <SEP> p <SEP> + <SEP> Q (ss
<tb><SEP> o
<tb><SEP> w <SEP> p
<tb> eitherF (p) = <SEP> 0 <SEP> (p <SEP> + È0Q <SEP>) <SEP> 2 <SEP> + ww02 <SEP> - <SEP> w)
<tb><SEP> if <SEP> on <SEP> asks <SEP>(<SEP>"o
<tb> if <SEP> on <SEP> asks <SEP> t <SEP> a =) Q <SEP> N <SEP> W0, <SEP> on <SEP> gets::
<tb><SEP> 4Q
<tb><SEP> w <SEP> 2 <SEP> = <SEP> p + a <SEP> - <SEP> a <SEP> I2
<tb><SEP> F (P) = <SEP> Q <SEP> t <SEP> Q <SEP> t "<SEP> 2 <SEP> w <SEP> 2 <SEP> w <SEP> 2 + w1
<tb><SEP> (p + a) <SEP> + <SEP> I <SEP> (p + a)
<Tb>
The z-transfer function of the equivalent digital filter is the z-transform of the impulse response of the analog filter (synthesis by the pulse invariance method).

The z-transform of this function is then, using the tables of transforms in z, for example those found in the book "The Digital Filters" by R. BOITE and H. LEICH ed. MASSON, page 19, 1980:

 In this formula T e is the period of the sampling signal.

In the example, the sample rate is four times the frequency of the chrominance subcarrier for the system
PAL so that the filter can be used in a color television camera digital encoder for use in PAL and SECAM systems. It is known that with such a sampling frequency the PAL color coder can have a very simple structure if the sampling times correspond to the zero crossings of the two quadrature subcarriers because, in this case, each sample then corresponds to a single subcarrier which simplifies the mixing of these subcarriers.

So:
Fe = 4 x 4.43361875 MHz = 17.734475 MHz
1 #s
either Te = 17.734475 (10)
F (z) then has the value:
1 - 0.0796 z-1
F (z) = 1.6831x
1 - 0.0998 z-1 + 0.90945 z-2 (11)
The z transform of N (p) is then
1 0.5941 (1 - 0.0998 z-1 + 0.90945 Z-2)
N (z) = F (z) = 1 - 0.0796 z-1 (12)
In this last formula the denominator is little different
of 1 because:
z -I = 1 and 0.0796 is small in front of 1.

In these conditions:
G (z) = 1 - 0.0998 z - + 0.90945 z -2 (14)
This transfer function is that of a non-recursive digital filter which has the advantage of being certainly stable.

This transfer function can still be written:
1 -1 -2
G (z) = 1.25 - 0.124775 z + 1.1368125 z) (15)
The coefficient of z is little different from 0.125 and the coefficient of z-2 is close to 1.125. These coefficients as well as the constant term 1.25 can therefore be realized by bit offsets because of the relations (3) mentioned above.

So that these offsets are towards the high-order bits, the lowest coefficient must thus be equal to 1, we choose a transfer function in z defined by the relation below:
H (z) = 10 - - + 9 -2 (16)
This transfer function corresponds, by a factor, to the function G (z) in which the coefficients have values given by the relations (3).

The coefficients of the function H (z) are easily realized because:
10 = 2 + 2
(17) 9 = 2 + 1 = 2 + 2 - 1
To obtain the structure of the digital filter, we start with the following relation in which S (z) is the output signal and E (z) is the input signal:
S (z) = H (z) E (z) = (23 + 21) E (z) - z ~ 1 E (z) + (23 + 21 - 1) z2 E (z), let S (z) = 2 (22 + 1) [E (z) + z'2 E (z)] - [z-E (z) + z ~ 2 E (Z) 0 (18)
The filter thus has the structure shown in FIGS.
and 4.

 The input signal E (z) is applied to the input of a set of D-type flip-flops on the clock inputs of each of which a signal at the sampling frequency F e is applied. Such an assembly 15 performs the function z-1 and constitutes a delay element.

 The signal E (z) is also applied to the first input 16 of an adder 17 having a second input 18 connected to the output of the element 15 via an identical element 19 on the clock input. from which the sampling signal is applied.

 The output of the element 15 is also connected to the first input 20 of the second adder 21 whose second input 22 is connected to the output of the element 19.

 The output 17a of the first adder 17 is connected directly to the first input 23 of a third adder 24 whose second input 25 is also connected to the output 17a but via a connection 26 - shown in FIG. shifted from 2 bits to the most significant, that is, multiplying by 4.

 The output 24a of the adder 24 is connected to the positive input (a) 27 of a subtractor 28 via a connection 29 making a shift of 1 bit towards the highs, that is to say a multiplication by 2.

 The negative input (b) 30 of the subtractor 28 is connected directly to the output of the adder 21.

On the output of the subtractor 28 we obtain the output signal
S (z).

 In the example shown in Figure 3 the parallel outputs 17a of the adder 17 are four bits while the parallel inputs 25 of the adder 24 are 6 bits. The connection 26 consists of: connecting the first bit, of least weight, of the outputs 17a to the third bit, representing the number 22, of the inputs 25; to connect the second, third and fourth bits of the inputs 17a, respectively to the fourth, fifth and sixth bits of the inputs 25 of the highest weight. With such a connection the signal on the inputs 25 is the quadruple of the signal on the outputs 17a.

 The connection 29 which is a shift of 1 bit (FIG. 4) towards the high order consists in connecting: the first bit, of least weight, of the outputs 24a of the adder 24 to the second bit of the inputs 27 of the subtractor 28, and the second, third, fourth, fifth and sixth bits of the outputs 24a respectively at the third, fourth, fifth, sixth and seventh bits of the seven-bit inputs 27. Such an offset ensures multiplication by the factor 2.

 The calculation has shown that a digital filter of the type shown in FIG. 2 and thus having the z-transfer function corresponding to formula (16) above has, between 3.9 and 4.75 MHz, a response virtually identical to that of the analog filter whose transfer function is represented by formula (4) above.

Indeed we see in Figure 5, in which the frequency f in
KHz has been plotted on the abscissa and the attenuation coefficient A on the ordinate, that the curve corresponding to the response of the analog filter deviates very little from the curve 36 calculated from the formula (16).

 This finding has also been confirmed by experience.

 FIG. 6 shows a digital SECAM encoder using digital anticlock filter 40 according to the invention.

 In this encoder the digital signals R - Y and B - Y (R = red signal, B = blue signal and Y = luminance signal) multiplexed are applied to the input 41 of a digital frequency modulator such as that described in the French patent application 81 13981 of July 17, 1981 for "variable frequency generator digital type" in the name of the applicant.

Such a variable frequency generator comprises a digital integrator providing on its output a signal representing the primitive function of the input signal and returning to a threshold value when it reaches a limit value, the output signal of the integrator thus representing the phase the desired periodic signal, the desired signal then being applied to the input of a periodic phaselsignal conversion means for example read-only memory.

 The output signal 42 of the modulator 43 is applied to the input of the filter 40 whose output is connected to the first input 44 of a digital adder 45. The second input 46 of this adder receives the digital luminance signal Y. The output of the adder 45 is connected to the output 47 of the encoder via a digital-to-analog converter 48.

 FIG. 7 is a diagram of a digital filter whose z-transfer function is that defined by formula (5) above.

 The input signal E (z) is applied to the first input of an adder 50 whose output is connected to the positive input (a) of a subtractor 51 having its output connected to the input of a first delay element 52 performing the function -1 and synchronized to the sampling frequency. The output of the subtracter 51 is also connected to the positive input (a) of a second subtractor 53 whose output is itself connected to the positive input (a) of a third subtractor 54. The output of the subtracter 54 is connected to the first input of an adder 55 whose output is connected to the positive input (a) of another subtractor 56. The output of the subtracter 56 is connected to the first input of an output adder 57.

 The output of the delay element 52 is connected to the second input of the adder 55 via another delay element 58 as well as to the negative input (b) of the subtractor 53 via connections 59 performing a shift of - 4 bits.

This output of the connections 59 is connected to the negative input (b) of the subtractor 54 via other connections 60 performing a shift of -1 bit.

 The output of the delay element 52 is also connected to the second input of the input adder 50 through connections 61 performing a 4 bit shift.

The output of the second delay element 58 is connected on the one hand to the negative input (b) of the subtractor 51 via connections 62 effecting a shift of -2 bits and secondly to
the negative input (b) of the subtractor 56 via
connections 63 performing a shift of - 3 bits.

Finally the output of the connections 63 is connected to the second input
of the adder 57 via other connections 64
performing a shift of - 2 bits.

In this filter the coefficients A, B, C and D of the formula (5)
have the following values: A = 0.09375 = 2 4 +
B = 0.90625 = 1 - 2-3 +
C = 0.0625 =
D = 0.25 = 2 (19).

Claims (9)

 Pre-emphasis filter, of the anticlockwise type, for the chrominance signal of a SECAM encoder, characterized in that it is sampled, preferably digital, with a z-transfer function of the form, to a factor:  -1 -2  G (z) = a + bz + cz  a, b and c being constants.  2. Filter according to claim 1, characterized in that its z-transfer function is, by a factor:  G (z) = 1 - 0.0998 z-1 + 0.90945 z-2.  3. Filter according to claim 1, characterized in that its z-transfer function is, by a factor:  H (z) = 10 - z-1 + 9 z-2  4. Filter according to any one of the preceding claims, characterized in that the sampling frequency is four times the frequency of the chrominance subcarrier in the PAL color television system.  5. Filter according to claim 3, characterized in that it comprises: a first delay element (15) whose input is connected to that of the filter and whose output is connected, on the one hand, to the first input (20) of a first adder (21) and secondly at the first input (18) of a second adder (17) via a second delay element (19), the second input (17) 22) of the first adder (21) being connected to the output of the second delay element (19) and the second input (16) of the second adder (17) being connected to the input of the filter; a third adder (24) whose first input (23) is connected to the output of the second adder (17) and whose second input (25) is connected to said output (17a) of the second adder via a multiplier (26) by four; and a subtractor (28) whose positive input is connected to the output (24a) of the third adder (24) via a multiplier (29) by two and whose negative input (30) is connected to the output of the first adder (21), the output of this subtracter being connected to the output of the filter.  6. The filter of claim 3 or claim 5, characterized in that the digital coefficients are made by bit shifted connections (26, 29).  7. SECAM encoder characterized in that it comprises an anticlock filter (40) according to any one of the preceding claims.  Encoder according to claim 7, characterized in that the anticlock filter (40) is preceded by a modulator (43) comprising a digital type variable frequency generator having an input on which is applied a digital signal representing the desired frequency for the output signal and which comprises a digital integrator providing on its output a signal representing both the primitive function of the input signal and returning to a threshold value when it has reached a limit value.  9. Color television camera, characterized in that it comprises an encoder according to claim 7 or 8.