GB2066615A - Improvements to colour television decoding apparatus - Google Patents

Abstract

A discriminator circuit for detecting horizontal colour boundaries in picture areas of a colour television picture to be formed from a colour television signal incorporating quadrature modulated chrominance information comprises first comparator means (106) for comparing chrominance components of parts of the video signal relatively delayed by an even number of line periods, and a second comparator (135) for comparing the chrominance components of parts of the video signal relatively delayed by an odd number of line periods. Each comparator means is associated with means (101, 102 or 140) for cancelling the phase difference between the chrominance sub-carriers of the respective signals between which a comparison is to be made. The output from the discriminator circuit may be used to control comb filtering of the video signal to remove chrominance information from the main luminance signal by providing that the comb filter is disabled when a signal output occurs from either one of the two comparators (106, 135). Alternative filtering means, for example a bandpass filter may then be used in place of the comb filter in picture areas where the use of the latter would be inappropriate. <IMAGE>

Description

SPECIFICATION Improvements to colour television decoding apparatus This invention relates to colour television decoding apparatus, and is concerned with means for use in separating the luminance signal from a colour television signal. In particular, the invention concerns means for analyzing a television signal in order to detect when comb filtering of the signal will be ineffective in providing accurate separation of the luminance signal.

As is well known, a PAL or NTSC colour television signal comprises a luminance (Y) signal, and two colour difference or chrominance signals (U and V), which are quadrature modulated onto a sub-carrier near the upper end of the bandwith of the video signal. Difficulties exist, however, in processing such a video signal to remove the chrominance signal components whilst leaving the luminance signal of adequate bandwidth and without distortion by residual chrominance components. Bandpass filtering of the signal to remove chrominance information also removes high frequency luminance components, thus causing deteriorated definition in the television picture, whereas insufficient removal of the chrominance components causes so-called "cross-colour" distortion, which shows as coloured interference on high frequency luminance detail.

In order to overcome the above disadvantages, it has previously been proposed to filter the video signal by means of a so-called comb filter, wherein signals relatively time displaced by one or more line periods are combined to effect removal of the chrominance information in summing or subtraction circuits. This technique requires that the chroniinance information be averaged over two or more line periods, and thus leads to deteriorated performance at horizontal colour boundaries.

In order to overcome this latter disadvantage, it has hitherto been proposed to provide a colour television apparatus with alternative means of filtering the luminance and chrominance signals, in addition to a comb filter, and to effect selective actuation of the two alternative filtering means in accordance with an analysis of the picture content to determine the presence or absence of horizontal colour boundaries. Such an arrangement would, in theory, provide a best possible compromise solution, wherein the relative advantages of bandpass and comb filtering respectively are made use of in appropriate areas of the television picture.In practice, however, such an analysis of picture content requires a determination of whether the chrominance signals on two adjacent lines are substantially identical in the region of the picture under consideration, and this presents considerable problems due to the changing phase relationship of the chrominance components of the signal on alternate lines.

It is accordingly an object of the present invention to provide an improved means for analyzing the content of a television signal in order to enable selective actuation of a comb filter means in accordance with the content of the video signal.

In accordance with the present invention there is provided a discriminator circuit for processing a colour television signal containing quadrature modulated chrominance signals to detect the presence of horizontal colour boundaries in the television picture comprising means for providing from an input video signal, a first signal delayed by an even number of line periods, a second signal delayed by an odd number of line periods, and a third, substantially undelayed signal, a first comparator means for comparing the chrominance components of said first and third signals and providing an output signal when a difference occurs between said chrominance components, a second comparator means for comparing said chrominance components of said second and said first and/or third signals and providing an output signal when a difference occurs between said chrominance components, and means associated with or incorporated in each of said comparator means for cancelling the phase difference between the chrominance subcarrier signals of the respective signals to be compared, whereby in each case an output signal from the comparator is representative of a horizontal colour boundary in the television picture.

Since, in accordance with the invention, comparison of the chrominance components of adjacent lines of the television picture is effected in a given picture area by means of two separate sensing circuits which respectively act on signals which are relatively delayed by an odd and even number of line periods respectively, the possible errors caused in each separate sensing circuit due to the cancellation of the respective chrominance sub-carrier phase shift, are cancelled out. Thus if one sensing circuit should be ineffective in detecting a colour boundary represented by a relative phase shift in the chrominance signals which has been cancelled by the detecting circuit, the other one of the circuits will respond to the appropriate phase shift and provide an output signal.

Cancellation of the respective 900 and 1 800 phase shifts over the odd or even numbered line periods, where appropriate, may be effected by a variety of means.

Thus, for example, in the case of a 1800 phase shift the respective chrominance signals may be simply full wave rectified, whereas a 90" phase shift may be cancelled by appropriate synchronous demodulation of the chrominance signals to provide separate phase-corrected demodulated U and V signals for comparison.

Such synchronous demodulation may incorporate compensation for the V phase reversal of a PAL signal in known manner. Alternatively, with appropriate comb filtering to provide filtered chrominance signals, the 900 phase shift and V phase reversal may be cancelled by a known PAL modifying circuit, which then enables direct comparison of the modulated chrominance signals.

In accordance with a further, preferred feature of the invention, means may be provided for detecting high frequency luminance components of the video signal and inhibiting the effect of the discriminator circuit in the presence of high frequency luminance signal components, so that the associated comb filter means for providing the final luminance signal is retained in effect over parts of the video signal containing high frequency luminance information. This arrangement takes advantage of the observation that has been made, following examination of average picture content, that high frequency luminance components and horizontal colour transitions do not normally occur together.

The invention will now be described by way of example with reference to the accompanying drawings, in which: Figure 1 is a diagram of the video spectrum of a PAL television signal, Figure 2 is a block schematic diagram of a simple PAL signal decoder circuit, Figure 3 shows a part of the spectrum of Figure 1 in a shared part of the band, that is it includes luminance Y signals and chrominance signals U and V, to an expanded scale, Figure 4 is a block schematic diagram of a known delay line PAL signal decoder circuit, Figurer 5 is a block schematic diagram of another known delay line PAL signal decoder circuit with a modifier, Figure 6 is a block schematic diagram of a known two delay line PAL signal decoder with a modifier, Figure 7 is a block schematic diagram of a known four delay line PAL signal decoder, Figure 8 is a block schematic diagram of one embodiment of a two delay line PAL signal decoder, Figure 9 is a block schematic diagram of another embodiment of a PAL signal decoder, Figure 10 is a circuit diagram of one section of a comb fail detector according to the invention, Figure 11 is a similar view of one section of another embodiment of the invention, Figure 12 is a circuit diagram of a second section of a comb fail detector in accordance with the invention, which may be combined with that of Figure 10or11, Figure 13 is a circuit diagram of a second section of a further embodiment of comb fail detector in accordance with the invention, which may be combined with that of Figure 10 or 11, Figure 1 4 is a circuit diagram of a second section of a further embodiment of comb fail detector in accordance with the invention, which may be combined with that of Figure 10 or Figure 11, Figure 1 5 is a complete circuit diagram of a decoder apparatus incorporating a comb fail detector in accordance with the invention, Figure 1 6 is a circuit diagram illustrating the - implementation of the colour adaption provided by the circuit of figure 1 5, in a decoder apparatus, and Figure 17 is a circuit diagram illustrating a possible modification of the circuit of Figure 1 6 for further improvement of the performance of the decoder.

In the various schematic diagrams, like parts are given like references and it will be understood that only that part of a PAL television receiver circuit is shown which is considered necessary for an understanding of the invention, the reader being referred to any of the standard textbooks on television for a complete understanding of the system.

Figure 1 shows the signal spectrum for a PAL signal and it will be seen that the luminance signal 10 bandwidth which is about 5 or 5.5 MHz overlaps that of the chrominance signal 1 2 which has a bandwidth of about 2.6 MHz centered on a sub-carrier frequency of about 4.43 MHz. The luminance signal is obtained by adding predetermined amounts of the signals due to the Red, Green and Blue R, G and B respectively, components of the picture to be transmitted which will provide a monochrome picture, that is it should not provide at the receiver any colour (hue) information. The luminance Y may thus be expressed as a function (R + G + B) and usually as voltage E magnitudes, such as Ev = f (ER + EG + EB).

The chrominance signal is derived as two equal bandwith colour difference signals V = f (R - Y) and U = f (B - Y) and as Y includes all three primary colours it is then possible to derive R, B and G. In the PAL system the phase of the V signal is reversed every alternate line to reduce any changes in hue due to phase errors between the decoder local oscillator and the received subcarrier.

In a conventional simple PAL decoder 14 such as shown in Figure 2 the PAL input signal 10 is applied to an input terminal 1 6 and the chrominance is separated by means of a band pass filter 20 having a bandwith of about 3MHz centred on the sub-carrier and a notch or band stop filter 1 8 removes most of the chrominance energy from the luminance component. In both cases separation is not complete and the luminance signal suffers from a loss of high frequency components, and also from so called "cross-luminance" due to residual chrominance which shows as bands of crawling dots at sharp colour transitions. Similarly, the band pass filter 20 in the chrominance channel also passes high frequency luminance components giving rise to so called "cross-colour", which shows as coloured interference on high frequency luminance detail.

The two colour difference signals U and V are separated by synchronous demodulation in synchronous demodulators 22, 24 using appropriately quadrature phased reference subcarriers 2sinwt and +2coswt respectively.

The outputs U and V of the demodulators 22 and 24 are coupled through associated low pass filters 26 and 28 to output terminals 30 and 32 respectively. The Y signal appearing at terminal 34 is used to provide the luminance signal and the Y output together with the U and V colour difference signals are applied to a suitable matrix circuit where the signals R, G and B are derived. The matrix circuit is not shown but can be obtained from a text book on colour television such as one entitled "Colour Television with Particular Reference to the PAL System" by G. N. Patchett published by Norman Price (Publishers) Ltd. This schematic, and all others in this Specification, assume for simplicity that all circuit blocks have zero delay except for delay elements. In practice additional delays may be required to compensate for inherent delays in circuit blocks.

The PAL signal can be expressed by the following equations: En = E, + Eusin wt + Ecos wt En+l = Ey = Eusin wt -- E,cos wt w fsc = -- = f(283=3/4)=2SHz 27 where En and Yen+1 are coded signals on successive lines.

Ey is luminance component derived from the sum of fractions of the colour components, for example Ey = xER + yEG = zE, Eu and Ev are colour difference components derived from a function EB - Ey and a function ER - E Ev fsc is sub-carrier frequency f, is line repetition frequency and B EB and G can be derived from E E and E ER, E and E V, U V' The equations reveal two special properties of the PAL signal, firstly the phase reversal of the V signal on successive lines, and secondly the rather complex quarter line frequency offset relationship between sub-carrier and line frequencies.For the following description, where adjacent line relationships are to be considered, the 25Hz offset, which introduces a phase reversal of the sub-carrier once per field, may be ignored.

A close examination of the frequency spectrum in the shared part of the band shows that the energy peaks for the luminance and chrominance occur at different frequencies, as shown in Figure 3. The luminance energy occurs at harmonics of line repetition frequency and chrominance is offset by the quarter line relationship. The energy pattern indicates that better separation between luminance and chrominance could be obtained by using filters that have pass characteristics which 6dctir in peakatfiroughout the band. Such filters can be realised by combining two signals that have been delayed with respect to each other.If the signals are subtracted, then frequencies for which the delay represents complete cycles will be cancelled, but pass characteristics remain at other frequencies. the complementary filter is obtained by adding the two signals. The form of filter is commonly called a spatial or comb filter.

A simple realisation of the comb filter is used widely in conventional delay line PAL decoders such as the decoder 38 of Figure 4 to separate the U and V components. The delay length must be the nearest integral number of full or half subcarrier cycles in a line, i.e. 284 or 283+, and is normally implemented as a single narrow band delay of 283 cycles.

As illustrated, delay lines 40 and 42 giving delays of 283+ and 1/4 cycles respectively are used in the combining paths. The vector diagrams of the U and V signals are shown at selected points in the circuit to show their phases. This form has direct equivalence to a single delay, and is used here to aid comparison with later schematics where the equivalence is used for simplication. The sum and the difference of half of the magnitudes of the output signals from delay lines 40 and 42 are taken in adder and subtractor circuits 44 and 46 respectively to obtain the V and U signals. In addition to UIV separation, this filter averages hue errors and reduces unwanted luminance components, which in turn reduces "cross-colour" by 3dB.

The final U and V signals are effectively the average over two lines of U and V signals and are spatially displaced from the luminance signal in the vertical direction by half a line pitch, causing a small misregistration of the colour information.

The reversal of V phase on alternate lines, a feature of the PAL system designed to reduce hue errors, creates difficulties when implementing comb techniques between adjacent lines.

Fortunately a technique of so called PAL modification, disclosed by Bruch in June 1 966 of Telefunken Zeitung, Selected Papers II, PAL a variant of the NTSC colour television system, exists which provides reversal of V phase without modifying the other components of the signal. The process involves modulating the chrominance signal with a carrier of twice the sub-carrier frequency. The lower sideband of the resulting output represents the original chrominance signal with reversed V phase, and the unwanted upper sideband is removed by filtering.The process is explained through the following equations assuming a particular phase for the twice subcarrier frequency:

For the more general case of any phase of twice sub-carrier frequency:- (Eu sin wt + Ey cos wt) (2 cos (wt + cup)) = E, sin (wt + 180 + cup) -- E, cos(wt + 180 + cP) + + Eu sin(3wt + ) + E,cos (3wt + Thus the modifier affords a technique of reversing the V phase and also giving a phase shift to the complete chrominance signal as a function of the twice sub-carrier frequency. This property is employed in some of the following schematics where an additional phase change is required.

Should the modifer input signal contain luminance components which are not phase coherent with the sub-carrier, these components can appear at the output as alias components, where frequencies from the unwanted upper sideband overlap the wanted luminance components. The alias may add to or subtract from the wanted luminance causing peaks or troughs in amplitude to 6dB. The effect is only disturbing for high energy high frequency components.

Comb filter techniques are only fully effective when the information is unchanged between successive lines,. i.e. in large areas of the picture.

At horizontal transitions of information the filters break down and unwanted "cross luminance" and "cross chrominance" may appear at these boundaries.

Figure 5 shows a decoder 50 using one delay line 40 and a modifier as disclosed by Auty, S.J., Read, D.C., and Roe, G.D. PAL Colour picture improvement using simple Analogue Comb Filters.

J.S.M.P.T.E. October 1978 Vol 87 pp677-681.

The U and V channels to outputs 30 and 32 are similar to that of Figure 4 with the addition of band pass filters 52 and 54 as shown. As with the conventional delay line decoder shown in Figure 4, the U and V signals are approximately the average of the U and V signals averaged over two lines and there is a vertical misregistration of luminance and colour difference signals and some loss of chrominance vertical resolution. The luminance is now combed to eliminate "cross-luminance" and to give full band response at the expense of alias components. This is achieved by applying the delayed output from the delay line 40 to a modifier comprising a synchronous demodulator 56 supplied with a modulating carrier at twice the sub-carrier frequency, that is with a carrier -- 2sin 2wt as aforementioned.The output of circuit 56 is applied through a band pass filter 58, which filters out the upper side band, to the subtract input of a subtractor circuit 60.

It will be noted that the U and V signals are in phase with those applied to the add input of the subtractor 60 so that the U, V signals from the immediately preceding line are subtracted from the U, V signals of the line under consideration, effectively to reduce the U, V contamination of the luminance signal that is to reduce the cross luminance. There is no loss of diagonal resolution.

The vertical misregistration of the above arrangement can be eliminated by employing a second delay line 62 as shown in the decoder of Figure 6. The chrominance signal to the modifier 56 and 58 and one component to the U/V separator is formed from the average of the input and 2H delay path in a subtractor 66 which also removes the luminance. By using two delay lines 40, 64 as shown the luminance signal (plus unwanted chrominance signals) is delayed one line when applied to the add input of subtractor 60. The chrominance signal applied to the subtrahend input of adder 60 is the average of the two chrominance signals in the lines immediately preceding and immediately succeeding the luminance line applied to subtractor 60.The resultant signal thus lies spatially registered to the 1 H delay path from which the main luminance component and the second component to the U/V separator are obtained. The complete filter still has some alias components which are zero at line and half line frequency harmonics. There is again no loss of diagonal resolution. There is no misregistration between luminance and colour difference signals and the circuit represents a good compromise between vertical chrominance resolution and reduction of "cross colour".

Another arrangement of a decoder 70 that does not use a modifier but provides a "combed" luminance output is shown in Figure 7. In this case contributions are taken across two two-line delays for which the chrominance is anti-phase, the first two-line delay comprising delay circuits 40, 72 and the second comprising circuits 74, 76. The chrominance and luminance signals from the input 1 6 and 4H delayed path from the output of delay circuit 40 are averaged by taking the sum of half of the magnitudes of each in adder circuit 80 and it will be noticed that this averaged signal at the output of adder circuit 80 is in phase with that at the input 1 6. These average signals from the output of adder circuit 80 are subtracted from the 2H (two line) delayed signal appearing at the junction of delay circuits 72 and 74 in subtractor circuit 82 resulting in a chrominance signal only as the chrominance signals are in anti-phase. The chrominance signal appearing at the output of subtractor circuit 82 is passed through a bandpass filter 84 to the subtract input of a subtractor circuit 86 the other input of which is coupled to the main luminance signal appearing at the junction of the delay circuit 72, 74. Thus the output of circuit 86 appearing at terminal Y is the luminance signal with the U and V effectively cancelled because they appear in phase at the circuit 86. The combed chrominance signals appearing at the output of circuit 84 are then synchronously demodulated for example as described with reference to Figure 2 to provide the U and V signals. The characteristics of this filter are such that it does not suffer from luminance alias components, and "cross luminance" is substantially eliminated.However, the vertical chrominance resolution is degraded by taking contributions over a five line aperture.

Another arrangement of filter 90 using only two delay lines 40 and 64 and no modifier is shown in Figure 8. In this case the signals from the input 1 6 undelayed path and the 2H delayed paths are summed in adder circuit 80, thus removing the antiphase chrominance signals. The resulting luminance only signal is subtracted in subtractor circuit 82 from the 1 H delayed path giving chrominance only, which is then passed through filter 84 and subtracted from the main luminance in subtractor 86 to give a combed luminance output at terminal 34. The combed chrominance is synchronously demodulated to U and V at terminals 30 and 32 as previously described with reference to Figure 2. This filter does not suffer from luminance alias components, and "cross luminance" is eliminated.There is however a loss of diagonal resolution with luminance zeros at subcarrier frequency and 78 cycles/picture height.

The filter retains full vertical chrominance resolution but suffers from U-V crosstalk when chrominance does not cancel in the adder. These spurious chrominance signals do not have the correct phase for proper detection in the synchronous demodulators. such crosstalk occurs mainly at the edges of coloured objects but "Cross colour" suppression is good.

It is clear that for all the comb filter arrangements so far discussed, no single circuit has optimum performance in respect of vertical and diagonal resolution, alias components and U/V crosstalk. All the circuits using more than one line delay however, result in vertical registration of the separated luminance and colour difference components, which is of great importance if the separate signals are to undergo further processing and re-encoding to a composite signal. The solution to derive an arrangement for a practical comb decoder is to combine two of the above arrangements to afford the best overall compromise in performance.

The final arrangement 94 is a combination of the two delay line circuit with modifier of Figure 6 and the two delay line circuit with the subtractor of Figure 8 to give the arrangement shown in Figure 9. The overall performance is an average of the two individual circuits. For the luminance signal, cross luminance is eliminated by averaging in an adder circuit 96 the output of the twice subcarrier frequency synchronous demodulaotr 56 (which provides the average of the chrominance signals of the lines immediately preceding and succeeding the main luminance line derived from the junction of delay circuits 40 and 64 and inphase therewith) and the output of the adder circuit 82 which provides a chrominance signal of the main luminance line. Thus the luminance output of adder 60 appears at terminal 34 with interference due to chrominance signals substantially eliminated.Alias components exist but are reduced by 6dB on the single circuit, and although there is a loss of diagonal resolution the zero now occurs at 1 56 cycle/picture height. The chrominance characteristics represent the best compromise between vertical resolution and cross colour suppression. Although there is some U-V cross talk it is reduced by 6dB on the single circuit level although it should be stated again that such cross-talk occurs only on colour transitions.

As stated earlier, comb filter techniques are only completely effective when there are minimal changes in information over the line aperture used for decoding and this circuit of Figure 9 is no exception. The breakdown of the filter is observed as cross-luminance and U-V cross-talk on horizontal transitions. In the description we have referred, where appropriate, to obtaining the average of various signals, for example in Figure 9 at adder circuit 96 the sum of half the magnitude of two signals is obtained. However, while preferably to obtain average values, some improvement would be obtained if signals other than half the value were used, for example, both input signals to an adder or other averaging circuit could be increased or decreased, or one signal could be increased and the other decreased.

Comb filters for decoding PAL signals, as described previously, have the disadvantage that failure occurs when there is a difference in the information on the successive lines over which the comb filter aperture operates. This failure is mainly observed at horizontal colour transitions.

This is observed as "cross luminance" in the luminance channel in the form of horizontal dots of colour sub-carrier which may be of larger amplitude than the sub-carrier level on the equivalent single line. Similarly, at failure, the chrominance channel may suffer from U/V crosstalk and "cross colour" is the disturbing luminance components fall within the chrominance band. Failure may also occur at horizontal transitions of high frequency luminance components due to propogation through the chrominance cancelling circuits.

To overcome the failure of the comb filters it is necessary firstly to determine a method for detecting all conditions of failure and secondly to modify the decoder operation to remove or reduce the unwanted components.

Although comb failure manifests itself in several ways, as outlined above, the root cause is the change in signal content on successive lines, and in particular the chrominance and line repetitive high frequency luminance. Detection methods are therefore optimally based on examination of these components over the successive lines of the filter aperture.

In the PAL system there is a quarter line frequency offset relationship between sub-carrier and lire frequency. In addition, the V component phase is reversed on alternate lines. For areas of constant colour there is a 900 phase shift with V component reversal between alternate lines and a 1 800 shift with no reversal of V over a two line period.

An ideal comparator for detection would only produce an error signal when there is a chrominance or high frequency difference, and for areas of constant colour or line repetitive luminance there would be no error output.

Chrominance signals can be directly compared in modulated form if there is a like phase relationship and V component phase, for example over a five line aperture. If there is a sub-carrier phase change or V phase reversal, then the chrominance must be demodulated or modified to remove the changes before comparison.

The first class of detector operates over a two line period and compares modulated chrominance after removal of the 1 800 phase change. Figures 10 and 11 illustrate two possible implementations of this detector. In the detector 100 of Figure 10 the signals from the input and 2H delayed paths are band-pass filtered in filters 101,102 respectively to select the chrominance and high frequency luminance which are then full wave rectified in rectifiers 103,104 respectively to remove the 1 800 phase difference before final comparison in a subtractor 106.In the detector 108 of Figure 11, signals from the input and 2H delayed paths are added in adder 110 to cancel the 1 800 related chrominance and the output is an average luminance signal which is then subtracted in subtractor circuits 112, 11 4 frorn the two paths to give chrominance only signals.

The 1 800 phase difference is removed by subtracting in the final comparator 116. The disadvantage of this class of detector 100 or 108 is that in removing the 1800 phase difference the detector is now insensitive to horizontal transitions between two colours which are represented by a 1800 phase inversion of the subcarrier.

This particular transition can give rise to a high level of cross-luminance in a comb decoder and is therefore a severe limitation in the sole use of this class of detector.

A second class of detector operates over one line period and the 900 phase shift and V phase reversal is removed by synchronous demodulation, using suitably phased sub-carrier, before comparison. Figures 12 and 13 illustrate two possible implementations 11 8 and 120 of such a detector. In Figure 12 the average chrominance between the direct and 1 H delay paths, and the corresponding average between the 1 H and 2H delayed paths are obtained by subtraction in subtractors 1 22 and 124 respectively, then followed by demodulation in demodulator 126 to remove the 900 phase shift and V phase reversal before final comparison in separate U and V subtractors 1 28, 1 30 respectively.

In Figure 13 the two signals before demodulation in demodulator 131 are the average chrominance from the input and 2H delay paths, and the combed chrominance from the 1 H delayed path obtained from summing circuits 1 32 and 134,136 respectively. This class of detector 11 8 and 120 has good sensitivity for horizontal transitions of 1 800 chrominance phase change, however due to removal of the 900 phase shift there is insensitivity to certain transitions between colours represented by some 900 phase changes of cub-carrier. The detectors also suffer from the complexity of demodulation in each comparison path.

A further derivative 121 of this form of detector utilizes a PAL modifier 140 to implement a 900 phase change and V phase reversal. It is then possible to make the final comparison between modulated chrominance signals. Figure 14 shows the appropriate modifier used in place of the demodulators of Figure 1 3. Once again, however, this detector is insensitive to some transitions represented by 900 phase changes of sub-carrier.

Both classes of detector are sensitive to transitions between some changes in frequency of line repetitive high frequency luminance components.

It should be apparent that no single comb fail detector operating over three successive lines will be sensitive to all horizontal chrominance and high frequency luminance changes, and it is necessary, in accordance with the invention, to combine the two classes of detector to provide a complete detection of comb decoder failure. The particular choice of implementation depends on the configuration of the decoder since the comparison signals may already be present for decoding functions.

A proposed form of best compromise comb decoder 142 is illustrated in Figure 15 with the addition of failure detectors as previously discussed for Figure 10 and Figure 14. The latter implementation of a class two detector is used here since the the decoder 142 already incorporates a PAL modifier 140.The two outputs from the detectors 106, 135 are summed in adder 1 44 to give a single error signal. Adjustable level detector and time delay circuits are incorporated within the finial summer 144 so that error signals are only available for failures of succifient amplitude and time duration to cause subjectively annoying disturbances to the picture produced by applying the decoder outputs to a picture monitor.

Having met the requirement for detection of comb decoder failure it is necessary to consider the modifications or adaptations that are required to remove unwanted components due to failure. It is not possible to reconfigure the schematic to an alternative comb arrangement as any comb decoder takes contributions from successive input lines and will exhibit similar failure conditions at horizontal transitions. The comb decoder under consideration has the property that there is no vertical mis-registration between the separated luminance and colour difference signals. It is required to maintain this property during adaptation and the choice of reconfiguration is restricted to that of a simole PAL decoder. The procedure adopted is to modify the luminance channel to include a notch or equivalent chrominance band rejector circuit, and switch the 1 H delay signal direct to the chrominance demodulators. This adaption is shown schematically in Figure 1 6. For simplicity this diagram assumes that all blocks have zero delay except the delay lines. In practice this is not a valid assumption and inherent circuit delays may dictate a practical form of reconfiguration different to that shown in order to maintain phased timing at the demodulators and no change in delay to the separated outputs during adaption. Similarly the switches are shown schematically and in practice will be electronic switches with controlled rate switching so that the adaption can be switched during a part of a line without disturbance to the output signal.

It is worthwhile to consider the overall effect of adaption on the performance of the proposed comb decoder. At a horizontal colour transition the cross-luminance in the luminance channel is removed or considerably reduced, and similarly the U/V crosstalk is eliminated from the colour difference signals. The penalty for this improvement is the normal deficiencies of simple PAL decoding, namely, a loss of luminance detail in the luminance channel due to the notch action and the possibility of cross-colour occuring in the chrominance if high frequency luminance is present at the horizontal transition.It should be emphasized however that this penalty only occurs during adaption.a Adaption may also occur at horizontal transitions of line repetitive high frequency luminance where there is no colour transition, the penalty is again one of loss of luminance detail and the possibility of cross colour.

A further improvement may be included to reduce the possibility of cross-colour during adaption. Examination of average picture content indicates that high frequency iuminance components and horizontal colour transitions do not normally occur together. Thus cross-colour can be reduced by inhibiting the chrominance adaption if high frequency luminance is present during comb failure.

Figure 1 7 shows a simple arrangement to effect this improvement. High frequency components present in the luminance output Y are separated in a high pass filter 145 and peak rectified in detector 146. The detected level is referenced to a threshold level at the input to comparator 147 and output of which inhibits the ~chrominance adaption control via electronic switch 1 48.

Although the above discussions on comb -decoder failure have concentrated on the PAL system, the broad principles on which detection is based is applicable to the NTSC colour system.

The apparatus as described could be implemented using analogue or digital techniques or a combination of both.

Claims (11)

1. A discriminator circuit for processing a colour television signal containing quadratue modulated chrominance signals to detect the presence of horizontal colour boundaries in the television picture, comprising means for providing from an input video signal, a first signal delayed by an even number of line periods, a second signal delayed by an odd number of line periods, and a third, substantially undelayed signal, a first comparator means for comparing the chrominance components of said first and third signals and providing an output signal when a difference occurs between said chrominance components, a second comparator means for comparing said chrominance components of said second and said first and/or third signals and providing an output signal when a difference occurs between said chrominance components, and means associated with or incorporated in each of said comparator means for cancelling the phase difference between the chrominance sub-carrier signals of the respective signals to be compared, whereby in each case an output signal from the comparator is representative of a horizontal colour boundary in the television picture.

2. A circuit as claimed in Claim 1, wherein the arrangement is such that said fist signal is delayed with respect to said third signal by two line periods, and said second signal is delayed with respect to said third signal by one line period.

3. A circuit as claimed in Claim 1 or 2, further including means for bandpass filtering said first and third signals respectively to remove the low frequency luminance components from said first and third signals, leaving the relatively higher frequency, chrominance and luminance components, wherein said means for cancelling the phase difference between the chrominance sub-carrier signals of said first and third signals comprises means for full wave rectifying the respective output signals from said bandpass filtering means, and wherein said first comparator means comprises a subtraction circuit for subtracting one of said full wave rectified signals from the other.

4. A circuit as claimed in Claim 1 or 2, further including means for summing said first and third signals to provide a fourth signal comprising substantially the average of the luminance components of said first and third signals with the chrominance components thereof eliminated, means for subtracting said fourth signal from said first signal to provide a fifth signal comprising substantially only the chrominance component of said first signal, means for subtracting said fourth signal from said third signal to provide a sixth signal comprising substantially only the chrominance component of said third signal, and wherein said first comparator means comprises a subtraction circuit for subtracting one of said fifth and sixth signals from the other.

5 A circuit as claimed in any one of Claims 1-4, being adapted for use with a PAL video signal and further including means for subtracting one of said first and second signals from the other to provide a first difference signal comprising the difference between the modulated chrominance components of said first and second signals with the luminance component of said first and second signals substantially removed, means for subtracting one of said second and third signals from the other to provide a second difference signal comprising the difference between the modulated chrominance components of said second and third signals, means for synchronously demodulating said first difference signal to provide therefrom demodulated U and V chrominance signals, means for synchronously demodulating said second difference signal to provide therefrom demodulated U and V chrominance signals, the respective sub-carrier frequencies of said synchronous demodulators being so selected that the relative 90" phase shift and V phase reversal is removed from the respective demodulated signals, and wherein said first comparator means comprises two separate comparators for respectively comparing the two pairs of demodulated U and V chrominance signals.

6. A circuit as claimed in any one of Claims 1-4, being adapted for use with a PAL video signal and further including means for summing said first and third signals to provide a fourth signal comprising substantially the average of the luminance components of said first and third signals with the chrominance components thereof removed, means for subtracting said fourth signal from said second signal to provide a first comparison signal comprising substantially the chrominance component of said second signal with the luminance component removed, means for subtracting one of said first and third signals from the other to provide a second comparison signal comprising substantially the average of the chrominance components of said first and third signals with the luminance component removed, and means for synchonously demodulating said first comparison signal to provide therefrom demodulated U and V chrominance signals, means for synchronously demodulating said second comparison signal to provide therefrom demodulated U and V chrominance signals, the respective sub-carrier frequencies of said synchronous demodulators being so selected that the relative 900 phase shift and V phase reversal is removed from the respective demodulated signals, and wherein said first comparator means comprises two separate comparators for respectively comparing the two pairs of demoduiated U and V chrominance signals.

7. A circuit as claimed in any one of Claims 1-4, being adapted for use with a PAI video signal and further including means for summing said first and third signals to provide a fourth signal comprising substantially the average of the luminance components of said first and third signals with the chrominance components thereof removed, means for subtracting said fourth signal from said second signal to provide a first comparison signal comprising substantially the chrominance component of said second signal with the luminance component removed, means for subtracting one of said first and third signals from the other to provide a second comparison signal comprising substantially the average of the chrominance components of said first and thircl signals with the luminance component removed, and means for modifying one of said first and socond comparison signals to cancel the 900 phase shift and V phase reversal in relation to the other, the said second comparator means being arrange to receive said first and second comparison signals.

8. A colour television signal decoding apparatus incorporating a comb filter means for filtering chrominance components from a video signal to provide a luminance signal substantially without chrominance components, a bandpass filter means for filtering chrominance components from said video signal to provide a luminance signal substantially without chrominance components, selector switching means for selectively enabling said bandpass filter and disabling said comb filter, and vice versa, and a discriminator circuit as claimed in any one of Claims 1-7, the arrangement being such that said switching means is caused to disable said comb filter means and enable said bandpass filter means in response to an output signal from either one of said first and second comparator means.

9. An apparatus as claimed in Claim 8, further comprising means for detecting high frequency luminance components of said video signal, said detecting means being arranged, in response to sensing of high frequency luminance components, to inhibit the effect of said discriminator circuit, whereby said comb filter remains effective for picture areas containing high frequency luminance information.

1 0. A discriminator circuit substantially as described herein with reference to figure 10 or 11 taken in combination with any one of Figures 12 to 14.

11. A discriminator circuit substantially as described herein with reference to Figure 1 5.

1 2. A discriminator circuit as claimed in Claim 10 or 11, as modified by Figure 17.