WO1993009635A1

WO1993009635A1 - Improvements in television systems

Info

Publication number: WO1993009635A1
Application number: PCT/GB1992/001988
Authority: WO
Inventors: Michael George Croll; John Oliver Drewery; James Edward Easterbrook; Graham Alexander Thomas
Original assignee: British Broadcasting Corporation
Priority date: 1991-10-30
Filing date: 1992-10-30
Publication date: 1993-05-13
Also published as: AU2808192A; GB2275586A; CN1076577A; GB2275586B; GB9407408D0; JPH07500711A; GB9123004D0; EP0610280A1

Abstract

A PAL wide aspect ratio luminance signal is transmitted in letterbox format with a main signal component occupying the central part of the vertical scan and a helper signal occupying the top and bottom borders. The input signal is sampled at 4/3 x 2fsc and is split into the two parts so that low horizontal frequency components are transmitted at 2fsc in the main component and high horizontal frequency components, e.g. above fsc, in the borders. The border signals are multiplexed up to 2fsc and preferably are frequency inverted. The main signal occupies three-quarters of the number of lines and the borders comprise the remaining one quarter. The resultant 2fsc signal can be applied to the luminance input of a phase segregated Weston PAL decoder for transmission.

Description

IMPROVEMENTS IN TELEVISION SYSTEMS

BACKGROUND OF THE INVENTION

This invention relates to apparatus for and methods of transmitting and receiving television signals, and particularly to the transmission of improved-quality television signals compared with those provided by the present-day transmission systems.

The invention could be applied to the NTSC system or to other systems, but will be described for convenience with reference to the PAL system and more particularly with reference to a system using a 16:9 picture aspect ratio. The signal can be displayed on a special 16:9 aspect ratio receiver, or can be displayed on an existing 4:3 aspect ratio receiver with what is known as a

"letterbox" display.

In a letterbox display, illustrated in Figure 24, the wide aspect ratio image is displayed with full horizontal width in the vertically central part of the screen, and the top and bottom of the picture are blanked off to black. This subjective effect is sometimes seen already when wide-screen cinematographic films are transmitted by broadcast television.

It is already known in a 625/50/2:1 standard PAL

transmission system to transmit 432 active lines which are then displayed on a 4:3 aspect ratio receiver in letterbox format. To display the picture on a 16:9 aspect ratio receiver the picture can be expanded vertically by increasing the vertical scan of the raster. This however results in the lines being further apart on the screen, and becoming visible, which is quite the opposite of the effect desired, namely to improve the subjective image quality.

There is also less definition in the horizontal direction.

Another proposal includes black bars at the top and bottom of the picture to include additional information so as to give more vertical detail. Related to this, United Kingdom Patent

Application GB-A-2 213 342 describes the transmission of the picture in letterbox format on a reduced number of lines, with low-amplitude non-visible additional information being transmitted in blanked-off areas at the top or bottom of the picture. Other prior documents in this field include the following, united Kingdom Patent Application GB-A-2 186 165 describes a CRT (cathode ray tube) display device which displays wide aspect ratio signals on a conventional display by reducing the scanning amplitude of the scanning raster. This however requires a modification to the CRT circuitry of the existing receiver.

United Kingdom Patent Application GB-A-2 203 011 describes a transmitter which includes also a decoder like that at a receiver. The signal to be transmitted as a normal video signal is decoded, and the decoded signal compared with, the video input signal. Any differences, representing impairments introduced in the

coding/decoding operation, are then transmitted with the video signal, for example during the vertical blanking, and are used in the receiver to counteract the effect of the impairments.

United Kingdom Patent Application GB-A-2 238 202 describes the transmission of an HDTV (high, definition television) signal as a main signal and a complementary signal. The split is achieved by making the main signal occupy three-quarters of the total number of lines, e.g. 432 out of 576, and carry the bottom three-quarters of the spectrum of the vertical detail in the picture. The

complementary signal has one quarter of the number of lines and carries the remaining top quarter of the spectrum of the vertical detail.

SUMMARY OF THE INVENTION

The invention in its various aspects is defined in the independent claims below to which, reference should now be made.

Advantageous features are defined in the appendant claims.

In a preferred embodiment of the invention to be described in more detail below a PAL wide aspect ratio luminance signal is transmitted in letterbox format with a main signal component occupying the central part of the vertical scan and a helper signal occupying the top and bottom borders. The input signal is sampled at 4/3 × 2 fsc and is split into the two parts so that low

horizontal frequency components are transmitted at 2 fsc in the main component and high horizontal frequency components, e.g. above fsc, in the borders. The border signals are multiplexed up to 2 fsc and preferably are frequency inverted. The main signal occupies three-quarters of the number of lines and the borders comprise the remaining one quarter. The resultant 2 fsc signal can be applied to the luminance input of a phase segregated Weston PAL coder for transmission.

The system provides continuous spatial and temporal pass-bands for luminance and chrominance. This is achieved by virtue of the phase-segregated Weston Clean PAL coding system, which avoids the need to "carve holes" in the spatio-temporal luminance spectrum in order to accommodate the U and V subcarriers. The confinement of spectrum folding techniques to helper information in the border assists good compatibility.

The luminance pass-band can extend to theoretical limiting resolutions of 6.5MHz horizontally (referred to a 15,625KHz scan) and to 160 or 216 cycles per active picture height vertically for the camera and film modes respectively. The usable, alias free, resolution achieved for PAL systems B and G by the embodiment described is judged to be 6.3MHz and 150 or 190 c/ph for luminance, and 0.5MHz and 60c/ph for each of the chrominance components. The shape of the luminance passband is hexagonal to optimise the perceived spatial resolution by making the distance from the origin to the band edge more nearly isotropic for all spatial frequencies.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention will now be described in more detail, by way of example, with reference to the accompanying drawings, in which:

Figure 1 is a block diagram of an encoder in accordance with this invention as used at a transmitter;

Figure 2 is a diagram showing the vertical-horizontal response of the luminance channel in the encoder of Figure 1;

Figure 3 is a block diagram of a clean PAL assembler at the encoder (transmitter) and splitter at the decoder (receiver);

Figure 4 is a block diagram of the decoder in a receiver embodying the invention;

Figure 5 is a timing diagram showing the writing to and reading from memories in the decoder; Figure 6 is a more detailed block diagram of a PAL splitter for combined Y and U±V as in a receiver;

Figure 7 shows the horizontal sub-band filter in the receiver of Figure 4;

Figure 8 shows the 4/3 × 2 fsc to 4/3 × 4 fsc post filter in the receiver;

Figure 9 illustrates the post filter aperture-to-sample relationship in the receiver in two phase positions;

Figure 10 is a block diagram of the 4/3 × 4 fsc to 720 samples/line down-converter in the receiver;

Figure 11 is a block diagram of the interlace-to-sequential converter in the receiver for converting from 432/2:1 to 432/1:1;

Figure 12 is a block diagram of the line rate up-converter in the receiver;

Figure 13 is a block diagram of the U±V sample rate converter in the receiver of Figure 4;

Figure 14 is a block diagram of the U/V vertical-temporal post filter in the receiver;

Figure 15 is a block diagram of the U/V display-rate up-converter in the receiver;

Figure 16 is a plot illustrating the vertical-temporal frequency response for the concaternated

sequential-to-interlace-to-sequential conversion in the transmitter and receiver combined;

Figure 17 is a plot showing the frequency response of the channel equaliser in the receiver;

Figure 18 is a plot showing the overall response of the equalised transmission channel;

Figure 19 is a plot showing the luminance post-filter vertical-horizontal response;

Figure 20 is a plot showing the luminance

interlace-to-sequential converter response in the receiver;

Figure 21 is a plot showing the luminance line rate

up-converter response in the receiver;

Figure 22 is a plot showing the chrominance pre-filter response; Figure 23 is a plot showing the chrominance line rate up-converter response; and

Figure 24 illustrates a letterbox display on a conventional 4:3 aperture ratio screen.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT

Figure 1 shows a block diagram of an encoder 10 embodying the invention. The encoder has inputs 12, 14, 16, 18 which receive respectively signals at the following picture standards:

(a) 1250/50/2:1

(b) 625/50/2:1

(c) 625/25/1:1

(d) 625/50/1:1

Format (a) has 1152 active lines per picture and formats (b), (c) and (d) have 576 lines per picture. Format (c) will normally be derived from film running at 25 frames per second.

Only one input is active at any one time. Inputs 12, 14 and 16 are connected respectively to converters 20, 22 and 24 which convert inputs of formats (a), (b) and (c) to the common standard (d).

This standard (d) is a progressive scan, that is to say it is non-interlaced. This format can support good vertical resolution without vertical-temporal alias.

Conversion from 1250/50/2:1 is achieved in converter 20 using a 15 line by 5 field filter on the 1152 active lines in a line sequential grid. This filter is designed to make the best possible progressive scan picture, taking the camera, display and visual characteristics into account. Conversion from 625/50/2:1 is performed using an 8 line by 5 field filter on a 576 sequential grid. Conversion from 625/25/1:1 is performed by simple field insertion, and is used only in the film mode.

The selected one of the outputs from converters 20, 22 and 24 together with input 18 is processed by a vertical low-pass filter 26. This provides vertical down-filtering and line rate conversion from format (d) to 432/50/1:1, where 432 represents the number of active lines in the picture. This is performed by up-sampling (increasing the sample rate) by a factor of 3, then filtering with a 41 tap vertical filter, and finally down-sampling (reducing the sample rate) by a factor of 4. The output of the filter 26 is effectively 432/50/1:1 as noted above. The luminance output of the filter 26 is re-converted to interlace form by a

sequential-to-interlace converter 28 to produce a 432/50/2:1 signal.

The converter 28 also applies a vertical-temporal filter function so as to optimally pack the interlaced transmission spectrum. In this manner the static vertical resolution is enhanced at the expense of vertical dynamic resolution. This is a well-known technique, and assists in achieving acceptable vertical resolution when the picture is displayed in letterbox format.

All of these operations are carried out at the input horizontal sampling density (i.e. sampling rate). The input horizontal sampling density is 1440 samples per picture width for format (a) and 720 samples per picture width for formats (b), (c) and (d). The signal may be analogue or digital at this point;

The theoretical vertical-temporal pre-post filter product for the encoder when combined with a corresponding decoder is shown in Figure 16.

Reverting to Figure 1, the output of the converter 28 is then applied to a horizontal frequency converter 30 which increases the sampling density horizontally so that it is 4/3 × 4 fsc, i.e. 16/3 fsc, where fsc is the colour subcarrier frequency. The sampling frequency increases in the horizontal direction only. The output of the horizontal frequency converter is then applied to a vertical-horizontal pre-filter 32. The pre-filter 32 is a diagonal filter with a band edge at -right-angles to the line connecting the origin and "4/3 × 2 fsc" as shown in Figure 2 labelled "compressed 2 fsc" sampling frequency. For these purposes such a filter can be constructed at minimal hardware cost, since alternate tap weightings are zero. The output of filter 32 is then applied to a 2:1 down-sampler 34, which sub-samples the filtered signal at a sample rate of 4/3 × 2 fsc while preserving the half-line and half-picture offset. This introduces aliasing at the upper band edge. Aliasing arises where spectrum folding causes high frequency signal components to be reflected back to cause interference components within the signal bandwidth. This aliasing here does not impair compatibility with existing receivers because it appears only in the border lines, and it is cancelled for the enhanced 16:9 display by appropriate post-filtering in the decoder.

The 4/3 × 2 fsc data stream is then split by a sub-band splitter 36, into a first stream at 2 fsc carrying the lower signal frequencies up to fsc, and a second stream at 2/3 × fsc which carries the higher signal frequencies. The split is handled on a purely horizontal basis, that is the sub-band filters are purely horizontal resulting in a purely horizontal spectral split as shown by the vertical line in Figure 2. The first data stream containing 3/4 of the samples occupies the middle three-quarters of the picture of the signal as transmitted, that is the middle 432 lines, and the second data stream containing 1/4 of the samples occupies the top and bottom borders of the picture as transmitted, that is the remaining 144 lines, split between the top and bottom of the picture. The second data stream is multiplexed in a multiplexer 38 to give a sample rate of 2 fsc This multiplexing can be achieved on a sample-by-sample basis, or can be achieved simply by putting the samples from three lines sequentially into one transmitted line. The thus-obtained 2 fsc signal is frequency-inverted about fsc, to reduce visibility, and fed at the appropriate time in the scan by a selector 40 to a clean PAL assembler 42. The amplitude of this the border signal is reduced by scaling to ensure that it is not visible on a conventional display, and is given a mean level corresponding to black.

The clean PAL assembler 42 is a so-called phase-segregated Weston clean PAL coding system. The basic Weston system is described in United Kingdom Patent Specifications GB-A-2 044 577 and GB-A-2 113 037. The operation is discussed in more detail in "A Compatible Improved PAL System" , EBU Review, February 1986. A further development is described in International Patent Application number WO-A-92/10068, published 11 June 1992.

The chrominance output of filter 26 is applied to a vertical-temporal pre-filter 44, which takes advantage of the fact that the U and V carriers are centred at different temporal frequencies of one quarter and three quarters of the picture frequency (6.25 and 18.75Hz) to extend the static vertical

chrominance resolution. This is worthwhile for PAL system I which permits a horizontal chrominance resolution of approximately

60 c/apw (cycles per active picture width); the benefit is less justifiable for PAL systems B and G which, only permit approximately 30 c/apw.

The filtered chrominance signals U and V are then applied to sample rate converters 46, 48 respectively, to convert the sample rate to 4 fsc to simplify the remainder of the processing, and are then horizontally low pass filtered to fsc/2 by filters 50, 52 and phase corrected in phase correctors 54, 56 to ensure correct chrominance positioning when decoded by a conventional PAL decoder. The U and V signals are then combined in a combiner 58 to form U + V and U - V on alternate lines, in accordance with a switching signal received at an input 60.

The phase-segregated Weston clean PAL coding system provides cross-effect free channels for a luminance component sampled at 2 fsc and a chrominance component sampled at fsc. The Weston system is also transparent to the luminance component, and to the chrominance component up to a horizontal bandwidth (sample density) determined by the bandwidth, of the link. In principle the shape of these channels in frequency space can be varied at will, provided that the maximum sample rates are not exceeded. Hitherto the requirement to provide good compatible picture quality, however, has prohibited the spectrum folding that would be required to provide anything other than a rectangular spectrum. This constraint has been overcome in the present system by dividing the

vertical-horizontal spectrum as shown by the vertical locus in Figure 2.

In Figure 2 is shown a plot in vertical/horizontal frequency space, for the luminance signal, with vertical frequency in cycles per active picture height (c/aph or simply c/ph) on the Y axis and horizontal frequency in cycles per active picture width (c/apw) on the X-axis. The operation is thus that the signal spectrum is divided along the horizontal frequency axis about colour subcarrier frequency fsc by sub-band coding with a ratio of 3/4:1/4. The lower three quarters extends from the origin to fsc and is

transmitted in the letterbox area of the signal at a sample rate of 2 fsc. This signal contains no spectrum folding so the level of artifacts is determined solely by the sub-band filter design. The upper one quarter extends from fsc to 4/3 fsc. This area of spectrum has spectral folding applied to extend its spectral limit for purely horizontal frequencies (zero vertical frequency) of 337 c/apw, equivalent to 6.5MHz. Samples describing this part of the signal spectrum are interleaved and frequency inverted before transmission in the border lines, where the use of spectrum folding has little effect on compatibility.

A conventional 4:3 aspect ratio receiver displays the letterbox part of the signal in the manner shown in Figure 24, and ignores the border areas which contain low-level signals which it interprets as black. Thus the signal is compatible in that it can be displayed on an existing receiver. An enhanced 16:9 aspect ratio receiver fills the screen with the displayed picture, which is generated by taking the high detail information given in the borders and adding it back to the low (and medium) detail information contained in the letterbox part of the signal. Such a receiver will be described with reference to Figure 4 showing a decoder for such a receiver.

First, a block diagram of the clean PAL assembler 42 and a corresponding splitter 80 in the receiver is shown in Figure 3.

The clean PAL assembler is supplied, as illustrated in Figure 1, with luminance samples at 2 fsc describing the lower three-quarters of the supported spectrum of Figure 2 during the letterbox period, and the upper one quarter, spectrally folded and multiplexed, during the upper and lower border periods. The borders contain samples belonging only to the field in which they are transmitted. The chrominance signal exists only on 216 lines per field (432 lines per picture) so the chrominance channel is unoccupied during the border lines. This vacant capacity could be used for additional helper information or to carry auxiliary data. The luminance signal Y is received at terminal 82 and is already sampled at 2 fsc. It is fed both directly and via a one-line delay 84 to a transversal filter 86 having filter function F1, the output of which forms the luminance component of the encoded PAL signal. The chrominance -signal is received at terminal 88 and is already modulated at fsc by a modulator 62 (Figure 1) and is then fed both directly and via a one-line delay 90 to a transversal filter 92 having filter function F2, the output of which forms the chrominance components. The outputs of these two filters are summed in a combining circuit 94 to form the PAL signal transmitted over the transmission link 100.

The encoder processing is assumed to be at 4 fsc in which case the modulations reduce to simple zero insertions and/or multiplying by ±1. This is by no means essential, however; the encoder and decoder could be implemented at any sufficiently-high line locked sample rate. The choice of subcarrier-related sampling merely reduces instrumental complexity, and the derivation of colour burst locked sampling clocks in the receiver is, if anything, more rugged and stable than for line locked clocks.

As shown in Figure 3, the splitter 60 and the receiver likewise incorporates two transversal filters 102 and 104 each fed from the transmission link 100 both directly and through a one-line delay 106. The filters 102 and 104 have filter functions F3 and F4 respectively, and provide as outputs a luminance signal from filter 102 and a chrominance signal from filter 104 comprising U + V and U - V on alternate lines.

The functions F1, F2, F3 and F4 in Figure 3 can be

determined from the above-mentioned documents describing the Weston PAL system. The Weston PAL system is closely related to the principles of sub-band analysis and synthesis in that the clean PAL assembler 42 operates as a sub-band decoder comprising two synthesis filters, and the splitter 60 operates as a sub-band encoder

comprising two analysis filters.

An enhanced receiver uses .a decoder 120 of the type shown in Figure 4, a timing diagram for which is shown as Figure 5. Referring to Figure 4, an incoming PAL signal is received at a PAL input 122 digitised (or at least sampled) at 4 fsc, and is passed through a transversal filter 124 to equalise the frequency response of the transmission channel 100. This transversal filter preferably has 27 terms and is symmetrical and uses a conventional folded accumulation ladder structure. Its coefficients may, for example, be as specified in Table 1 below, and its characateristics as shown in Figures 17 and 18. Figure 17 illustrates the channel equaliser frequency spectrum plotted against horizontal frequency in cycles per active picture width (c/apw), and Figure 18 illustrates the overall response of the thus-equalised channel. Moreover, the 4 fsc PAL is sampled at a specified phase namely at 45 degrees to the U sub-carrier.

The output of the channel equaliser 124 is applied to two transversal filters comprising the PAL splitter 80, as shown in Figure 3. One filter generates the luminance samples Y while the other generates the chrominance samples U±V.

The filters typically have 77 terms each of two lines on a 4 fsc lattice and are symmetrical, which halves the number of coefficients required. The luminance output can be thought of as a signal at 4 fsc in which alternate samples (on a 2 fsc quincunxial lattice) are zero; in practice the zero samples are not generated, so the output rate from the luminance splitter is actually 2 fsc, giving a further halving of the number of coefficients required at any one time. Similarly the chrominance output can be thought of as samples at 4 fsc of which three out of four samples are zero; thus the chrominance signal requires a sample rate of fsc and provides a further two-fold reduction in the number of coefficients required at any one time.

A more detailed block diagram of the clean PAL splitter is shown in Figure 6. Figure 6 and the subsequent Figures 7, 8 and 9 to 15 show the construction of the illustrated circuits, largely based on transversal filters, using conventional indications of circuit components in the form of sample delays, line delays, switches, multipliers, and adders and it would be totally

superfluous to the skilled readers of this specification to provide a textual description of what is more elegantly shown to them by these figures. Detailed textual description of these figures is thus unnecessary and therefore omitted.

The border samples are frequency inverted by multiplying by fsc in a multiplier 126, and passed through a look-up table 128 to undo any non-linearity applied at the encoder. The non-linearity is as specified in Table 7 below. The lines are then applied through a memory 130 to be stored in a re-circulating line memory 132 which holds them by means of a selector switch.134 for three lines in order tα undo the three-to-one line multiplexer performed at the encoder, assuming sample-by-sample multiplexing.

The letterbox samples are applied through a compensating and re-timing memory 136 to another selector switch 138 which selects the letterbox and border samples at appropriate parts of the scan and applies them to a horizontal sub-band filter 140 which performs sub-band decoding and requires the samples to be multiplexed in a ratio of three letterbox samples to one border sample. The 2 fsc letterbox samples are multiplexed with 2/3 × fsc samples from the border lines by simple switching by switch 138.

The memory 130 is a 72 line by 473 sample dual port field memory and provides vertical re-timing for the border lines. The memory 136 is a 144 line by 473 sample memory and provides re-timing for the letterbox lines, as shown in Figure 5. The resulting signal contains 216 lines per field but extending over the whole vertical scan period. This avoids the need for large quantities of storage in the subsequent line rate up-converter 148, but

complicates the description of sampling frequencies used throughout the decoder description. The term fsc when used in the decoder description should therefore be considered as a spatial description with respect to the raster lines (e.g. 283.7516 samples per line) rather than a one-dimensional sampling frequency.

Subsequent to the letterbox combiner 140 are arranged in series a post filter 142, a sample rate converter 144, an

interlace-to-sequential converter 146, and the line-rate

up-converter 148. The chrominance signal U±V from the PAL splitter 80 exists only during the letterbox lines, but must eventually be displayed over the full vertical scan period and with a line period of 32 microseconds. A 144 line by 235 sample dual port field memory 150 is used to spread the chrominance lines over the vertical scan period, with three output lines spaced by 64 microseconds followed by a vacant line, in a similar manner as used for the luminance. The only difference is the positioning of the line period

contraction which, in the chrominance, can be performed by the chrominance re-timing memory without exceeding the maximum

permissible sample rate later in the decoder. The same caution as just noted above for the luminance signal is required when

interpreting the meaning of subcarrier frequency.

The output of the memory 150 is then applied through a sample rate converter 152, a U/V demultiplexer 154, a

vertical-temporal post filter 156, and a line rate up-converter 158, to provide U and V output signals.

The individual component circuits will now be described in more detail.

The multiplexed letterbox and border samples, describing the lower and upper sub-bands of Figure 2, are re-combined in the horizontal sub-band filter 140, a block diagram of which is shown in Figure 7. The structure thereof is self-evident from the figure. The sampling density of the filter input and output is

4/3 x 2 fsc, which has a half line offset (i.e. a quincunxial sampling structure). The filter implements simultaneously the functions of a low-pass sub-band synthesis filter operating on the letterbox samples, and a high-pass synthesis filter operating on the border samples; the resultant signals are summed. The encoder sub-band filter and decoder sub-band filter are reciprocal

operations and are transparent when concatenated.

The output data from the filter 140 is at a sample density of 4/3 x 2 fsc," this is a quincunxial structure which is just adequate to support the signal spectrum but could not be used directly for display because of aliasing from the repeat spectrum at the upper band edge. The post-filter 142 performs a sample density increase to 4/3 x 4 fsc; this structure is orthogonal and over-sampled which places the alias from the spectral repeat out of band. The post-filter circuit is shown in Figure 8, and again is self-evident. It has a spatial frequency response as shown in Figure 19, its coefficient to sample relationship is shown in Figure 9, and its specification is given in Table 2 below.

The post-filter output is oversampled horizontally and, being fsc related, is only pseudo line locked. A truly line-locked sampling structure is required by the consumer electronics industry for use in displays. Consequently, a horizontal sample rate conversion is performed by a four-tap 256 phase sample rate converter 144 to 720 samples per active line, line locked. A block diagram of the sample rate converter is shown in Figure 10.

The signal at this point is fully restored with a near- minimal sample density both horizontally and vertically. The horizontal post filtering is performed by an analogue filter in the output digital-to-analogue converter whose performance can be accurately specified. Vertical post filtering however is normally performed by the display spot profile; this is a less than perfect operation and, for an enhanced display, necessitates conversion to a higher line rate to permit electronic vertical post-filtering. In this way the response of the electronic post-filter is dominant and signal frequencies that would give rise to dominant static or dynamic vertical aliasing are attenuated.

Luminance display rate up-conversion is optional but if included is performed in two parts. The first is interlaced-to-sequential conversion from 432/2:1 to 432/1:1 in converter 146 and uses inter-field filtering in order to reclaim the vertical

frequencies above 108 cycles per picture height that were placed on a 25Hz carrier by the encoder sequential to interlace converter.

The second part is line rate up conversion from 432 to 576 active lines in converter 148 to reduce the static vertical alias remaining after post filtering by the inevitably less than perfect display spot profile. Block diagrams of the interlace to sequential converter and line rate up converter are shown in Figures 11 and 12 respectively, and their frequency responses and coefficient patterns are given in Figure 20 and Table 3 and Figure 21 and Table 4

respectively. Figure 16 illustrates the vertical-temporal frequency response for the concaternated

sequential-to-interlace-to-sequential conversion in the transmitter and receiver combined.

The 216 lines per field period at the input to the interlace to sequential converter 146 appear in the time occupied by 288 normal television lines; the line period up to this point can be maintained at 64 microseconds with a 21 ⅓ microsecond gap or increased to 85 Vs microseconds. This is an arbitrary choice and, since the first operation performed by the converter is to contract the line period to 32 microseconds commensurate with display at 31.25 KHz, makes no difference to hardware complexity, delay capacities etc. The contraction is performed by a pair of dual port line memories, one of which is written into while the other is read. Note that the interpolator is active for three lines followed by one inactive line in which all data movements are suspended.

The interlace to sequential conversion is performed by a symmetrical vertical temporal filter in order to minimise

vertical-temporal artifacts e.g. 'cogging' on sloping lines.

The interpolator contains two signal paths, a direct path for output lines coincident with input lines and an interpolated path for the remaining output lines. The interpolation filter has four fixed taps in the current field which provide predominantly low vertical frequencies and five taps in the previous and succeeding fields which provide the predominantly high vertical frequencies, placed on a 25 Hz carrier by the encoder sequential-to-interlace converter. The direct and interpolated lines are fed in parallel, each at 720 samples per line, to the display line rate up converter.

The line rate up converter 148 operates by upsampling by a factor of three, applying a 15 tap vertical filter and downsampling by a factor of four. Thus the filter has four phases, three of which have four coefficients and the fourth has three coefficients. The input to the line rate up converter is consecutive sequential lines with the correct line period of 32 microseconds and at the correct frequency but with every fourth line missing. The lines required to contribute to the filter output are held in

recirculating dual port line memories with the latest input line being routed, as it appears, to the first delay whose contents are handed on to the next delay and so on. The switch sequence is therefore to accept new data for three lines then re-circulate for one line.

The chrominance signal at the output of the chrominance re-timing memory 150 is U±V at_a sample rate of fsc; this can be thought of the U component centred on the origin and the V component centred on 108 c/aph (referred to a 16:9 aspect ratio) and 12.5 Hz. The U component can therefore be separated by appropriate vertical- temporal post filtering. Similarly, applying the V axis switch places the V component at the origin and the U component at

108 c/aph and 12.5 Hz allowing separation of the V component.

The separation of the U and V is therefore a purely vertical-temporal operation; the horizontal chrominance signal bandwidth is about 0.6 Mhz for PAL systems B and G and approximately 1.1 MHz for system I, and can be fully supported by a horizontal sampling density of 180 samples per active line, i.e. CCIR REC 601 sampling density. Consequently, a sample rate conversion to 180 samples per line is performed on the chrominance at the output of the re-timing store and the direct and switched chrominance signals are multiplexed as U±V, V±U, U±V, V±U... to form a combined signal at a sampling density of 360 samples per line. This allows colour separation by vertical-temporal filtering and display line rate up conversion to be performed by the same hardware modules. The U±V sample rate converter is shown in Figure 13.

The vertical-temporal post filters for U and V separation are identical, which is convenient since they can be implemented by the same piece of hardware, and are temporally asymmetrical. The low vertical frequency components are derived from the earlier field, i.e. after the field delay and the high, frequency components from the later field. The filter aperture is slightly larger than for the luminance with, seven terms in the later field and six in the earlier. The memory control logic must take account of the one in four latency of the re-timed input signal. The filter block diagram is shown on Figure 14. The combined pre and post filter response is shown in Figure 22, and the post filter specification in Table 5. The output of the chrominance post filter is alternate U and V samples, each at 180 samples per active line, carried on 216 lines spread over the output active scan period with one line in every four being vacant. The chrominance display rate up-converter 158 is not required to support a vertical resolution above 108 cycles per picture height and is therefore a purely vertical interpolation process.

A block diagram of the display rate up converter 158 is shown in Figure 15 and its response and specification are given in Figure 23 and Table 6. It is similar to that used for luminance except that its input is multiplexed U, V, U, V and on an interlaced sampling lattice. The interpolator performs up sampling by a ratio of four, applies a 39 tap filter, and then down samples by a factor of three. There are therefore eight phases each having five coefficients. The line memories recirculate except when a new 32 microsecond input line is available.

The phase of the interpolator changes once per output line period but is the same for both U and V samples. The U and V signals can be separated by a simple switch to provide separate data streams at 180 samples per active line; the output is therefore in 4:1:1 format with luminance at 27 MHz.

It will be appreciated that the various filters etc. shown separately in the drawings may in fact be coalesced or arranged in other ways to provide the same overall input/output function. Some or all of the functions may be implemented in software.

In summary therefore the system, overall, starts by reducing the input to 432 lines at 4/3 × 4 fsc. The signal is then

pre-filtered using the overall function of Figure 2, and

down-sampled by 2 to give a sample rate of 4/3 × 2 fsc.

In accordance with the invention the signal is split into two parts by two filters, so that the main low horizontal frequency components are transmitted in the letterbox or main signal and the high horizontal frequency components are transmitted in the border regions. This is a so-called sub-band split; that is to say it does not matter that the filters are imperfect so long as they are complementary. The border signal is then re-arranged onto one third of the number of lines. It is highly preferable to frequency invert the signal, and use a companding function.

The letterbox and border lines are then re-combined. Each field contains in sequence 36 lines of border followed by 216 lines of main or letterbox signal followed by another 36 lines of border. This gives a total of 288 lines per field at a sample rate of 2 fsc. This sample rate is suitable to be applied to the luminance input of a phase segregated or Weston PAL coder.

While the description has been given in terms of a highly preferred 3 : 1 split, other ratios could be conceived. Equally some departure from 2 fsc as the mean sample rate is possible, though some loss of quality may result.

TABLE 1

Channel equaliser coefficients list

Filter coefficientsz y x in units of 1/512

0 0 -13 4.

0 0 -12 13.

0 0 -11 -46.

0 0 -10 56.

0 0 - 9 -15.

0 0 - 8 -51.

0 0 - 7 60.

0 0 - 6 12.

0 0 - 5 -74.

0 0 - 4 36.

0 0 - 3 59.

0 0 - 2 -72.

0 0 - 1 -22.

0 0 0 592.

0 0 1 -22.

0 0 2 -72.

0 0 3 59.

0 0 4 36.

0 0 5 -74.

0 0 6 12.

0 0 7 60.

0 0 8 -51.

0 0 9 -15.

0 0 10 56.

0 0 11 -46.

0 0 12 13.

0 0 13 4. TABLE 2

Luminance post- -filter coefficient list

Filter Coefficients z y x in units of 1/512

0 -2 -7 0.

0 -2 -6 -7.

0 -2 -5 0.

0 -2 -4 10.

0 -2 -3 0.

0 -2 -2 -8.

0 -2 -1 0.

0 -2 0 9.

0 -2 1 0.

0 -2 2 -8.

0 -2 3 0.

0 -2 4 10.

0 -2 5 0.

0 -2 6 -7.

0 -2 7 0.

0 -1 -7 0.

0 -1 -6 0.

0 -1 -5 0.

0 -1 -4 0.

0 -1 -3 0.

0 -1 -2 0.

0 -1 -1 0.

0 -1 0 0.

0 -1 1 0.

0 -1 2 0.

0 -1 3 0.

0 -1 4 0.

0 -1 5 0.

0 -1 6 0.

0 -1 7 0.

0 0 -7 -9.

0 0 -6 8.

0 0 -5 27.

0 0 -4 -9.

0 0 -3 -53.

0 0 -2 5.

0 0 -1 163.

0 0 0 250.

0 0 1 163.

0 0 2 5.

0 0 3 -53.

0 0 4 -9.

0 0 5 27.

0 0 6 8.

0 0 7 -9. Filter Coefficients z y x in units of 1/512

0 1 -7 0.

0 1 -6 0.

0 1 -5 0.

0 1 -4 0.

0 1 -3 0.

0 1 -2 0.

0 1 -1 0.

0 1 0 0.

0 1 1 0.

0 1 2 0.

0 1 3 0.

0 1 4 0.

0 1 5 0.

0 1 6 0,

0 1 7 0.

0 2 -7 0.

0 2 -6 -7.

0 2 -5 0.

0 2 -4 10.

0 2 -3 0.

0 2 -2 -8.

0 2 -1 0.

0 2 0 9.

0 2 1 0.

0 2 2 -8.

0 2 3 0.

0 2 4 10.

0 2 5 0.

0 2 6 -7.

0 2 7 0.

TABLE 3

Luminance interlace-to-sequential converter coefficient list

Filter Coefficients z y x in units of 1/512

1 -4 0 7.

1 -3 0 0.

-1 -2 0 -44.

-1 -1 0 0.

1 0 0 67.

1 1 0 0.

-1 2 0 -44.

1 3 0 0.

-1 4 0 7.

0 -4 0 0.

0 -3 0 -1.

0 -2 0 0.

0 -1 0 136.

0 0 0 256.

0 1 0 136.

0 2 0 0.

0 3 0 -1.

0 4 0 0.

1 -4 0 7.

1 -3 0 0.

1 -2 0 -44.

1 -1 0 0.

1 0 0 67.

1 1 0 0.

1 2 0 -44.

1 3 0 0.

1 4 0 7. TABLE 4

Luminance line rate up-converter coefficient list

Filter Coefficientsz y x in units of 1/512 0 -7 0 -6.

0 -6 0 -13.

0 -5 0 -14.

0 -4 0 0.

0 -3 0 33.

0 -2 0 76.

0 -1 0 115.

0 0 0 130.

0 1 0 115.

0 2 0 76.

0 3 0 33.

0 4 0 0.

0 5 0 -14.

0 6 0 -13.

0 7 0 -6.

TABLE 5

Chrominance vertical-temporal post-filter coefficient list

Filter Coefficients z y x in units of 1/512

-1 -6 0 0.

-1 -5 0 -56.

-1 -4 0 0.

-1 -3 0 -5.

-1 -2 0 0.

-1 -1 0 65.

-1 0 0 0.

-1 1 0 6.

-1 2 0 0.

-1 3 0 -46.

-1 4 0 0.

-1 5 0 -6.

-1 6 0 0.

0 -6 0 26.

0 -5 0 0.

0 -4 0 18.

0 -3 0 0.

0 -2 0 108.

0 -1 0 0.

0 0 0 228.

0 1 0 0.

0 2 0 168.

0 3 0 0.

0 4 0 14.

0 5 0 0.

0 6 0 -8.

TABLE 6

Chrominance line rate up-converter coefficient list

Filter Coefficients z y x in units of 1/512 0 -19 0 1.

0 -18 0 0.

0 -17 0 0.

0 -16 0 0.

0 -15 0 -1.

0 -14 0 -2.

0 -13 0 -4.

0 -12 0 -6.

0 -11 0 -7.

0 -10 0 -6.

0 -9 0 -4.

0 -8 0 0.

0 -7 0 7.

0 -6 0 16.

0 -5 0 26.

0 -4 0 37.

0 -3 0 48.

0 -2 0 57.

0 -1 0 62.

0 0 0 64.

0 1 0 62.

0 2 0 57.

0 3 0 48.

0 4 0 37.

0 5 0 26.

0 6 0 16.

0 7 0 7.

0 8 0 0.

0 9 0 -4.

0 10 0 -6.

0 11 0 -7.

0 12 0 -6.

0 13 0 -4.

0 14 0 -2.

0 15 0 -1.

0 16 0 0.

0 17 0 0.

0 18 0 0.

0 19 0 1.

Claims

CLAIMS:

1. A method of transmitting a luminance signal, comprising the steps of splitting the signal-into first and second parts, the first part comprising signal components having low horizontal frequencies and the second part representing signal components having high horizontal frequencies, and transmitting the first part on first selected lines of the transmitted signal and the second part on second selected lines of the signal.

2. A method according to claim 1, in which the first part comprises signal components below a predetermined frequency and the second part comprises frequency components above the predetermined frequency.

3. A method according to claim 2, in which the predetermined frequency is fsc.

4. A method according to claim 1, in which the first part comprises substantially three times the frequency components contained in the second part.

5. A method according to claim 1, in which the transmitted signal is applied to one input of a phase-segregated coder.

6. A method according to claim 5, in which the signal is a PAL-type signal and the phase-segregated coder is a Weston PAL coder.

7. A method according to claim 6, in which a chrominance signal is applied to the other input of the Weston PAL coder.

8. A method according to claim 1, in which the second part is compressed in time before combining with the first part to form the transmitted signal.

9. A method according to claim 1, in which the second part is frequency inverted before combining with the first part to form the transmitted signal.

10. A method according to claim 1, in which the first selected lines are central lines of the vertical scan and the second selected lines are top and bottom lines of the vertical scan.

11. A method of transmitting a luminance signal, comprising the steps of generating a sampled signal with a sampling frequency of 4/3 × 2 fsc, splitting the signal into first and second parts with the first and second parts containing different frequency components of the signal, the first part having a sampling frequency of 2 fsc and the second part having a sampling frequency of 2/3 fsc, compressing the second part at a frequency of 2 fsc, and

transmitting the first and second parts alternately in time.

12. A method according to claim 11, in which the sampled signal with a sampling frequency of 4/3 × 2 fsc is generated by generating a signal of 4/3 × 4 fsc and down-sampling to 4/3 × 2 fsc.

13. A method according to claim 11, in which the first part comprises signal components below fsc and the second part comprises frequency components above fsc.

14. A method according to claim 11, in which the transmitted 2 fsc signal is applied to one input of a phase-segregated coder.

15. A method according to claim 14, in which the signal is a PAL-type signal and the phase-segregated coder is a Weston PAL coder.

16. A method according to claim 15, in which a chrominance signal of sampling rate fsc is applied to the other input of the Weston PAL coder.

17. A method according to claim 11, in which the second part is compressed in time before combining with the first part to form the transmitted signal.

18. A method according to claim 11, in which the second part is frequency inverted before combining with the first part to form the transmitted signal.

19. A method according to claim 11, in which the first part is transmitted on central lines of the vertical scan and the second part is transmitted on top and bottom lines of the vertical scan.

20. Apparatus for transmitting a luminance signal, comprising splitting means for splitting the signal into first and second parts, the first part comprising signal components having low horizontal frequencies and the second part comprising signal components having high horizontal frequencies, and means for transmitting the first part as first selected lines of the

transmitted signal and the second part as second selected lines of the signal.

21. Apparatus for transmitting a luminance signal, comprising the steps of generating a sampled signal with a sampling frequency of 4/3 × 2 fsc, means for splitting the sampled signal into first and second parts containing different frequency components of the signal, the first part having a sampling frequency of 2 fsc and the second part having a sampling frequency of 2/3 fsc, means for compressing the second part at a frequency of 2 fsc, and means for transmitting the first and second parts alternately in time.

22. A method of receiving a transmitted luminance signal, comprising separating the lines of the transmitted signal such that first lines form a first part and second lines form a second part, the first part comprising signal components having low horizontal frequencies and the second part representing signal components having high horizontal frequencies, relatively re-timing the first and second parts, and recombining the first and second parts to form an output signal in which the low horizontal frequency components are formed from the first part and the high horizontal frequency components are formed from the second part.

23. A method of receiving a transmitted luminance signal having a sample rate of 2 fsc, comprising splitting the incoming samples in time so that groups of incoming samples form a first part and groups of incoming samples form a second part, expanding the second part to a frequency of 2/3 fsc, and recombining the first and second parts to form an output signal in which, the first and second parts constitute different frequency components of the signal, the output signal having a sampling rate of 4/3 × 2 fsc.

24. Apparatus for receiving a transmitted luminance signal, comprising means for separating the lines of the transmitted signal such that first lines form a first part and second lines form a second part, the first part comprising signal components having low horizontal frequencies and the second part representing signal components having high horizontal frequencies, means for relatively retiming the first and second parts, and means for recombining the first and second parts to form an output signal in which the low horizontal frequency components are formed from the first part and the high horizontal frequency components are formed from the second part.

25. Apparatus for receiving a transmitted luminance signal having a sample rate of 2 fsc, comprising means for splitting the incoming samples in time so that groups of incoming samples form a first part and groups of incoming samples form a second part, means for expanding the second part to a frequency of 2/3 fsc, and means for recombining the first and second parts to form an output signal in which the first and second parts constitute different frequency components of the signal, the output signal having a sampling rate of 4/3 × 2 fsc.