The present invention relates to the trans¬ mission and reception of T.V. video signals in digital orm.

One method of transmitting digital T.V. video signals is to convert analog T.V. video signals into digital signals by sampling the analogue signals at a frequency which is a multiple of the frequency of the video sub-carrier (fsc) in the case of a PAL signal. Each sample is converted to a digital word containing for example 8 bits. If the sampling frequency is -ϊfsc then when such a digital signal is serialised for transmission, the data rate is approximately l'i2 Mbit/s. This data-rate is excessive for many applications, and work has therefore been carried out by a number of organisations to reduce the data-rate. One technique (known as Sub-Nyquist sampling) involves the use of a digital filter which allows alternate words to be removed from the •_ fsc signal. This produces a 'Sub-Nyquist' signal with a word frequency of 2 fsc and a serial data-rate of approximately 71 Mbit/s. Applications exist which require a data-rate which is lower than 71 Mbit/s.

The present invention is based on the use of di erential pulse code modulation (DPCM) since such techniques will reduce the data rate. It is believed

that DPCM techniques are now well known in the art but a full description of the basis of DPCM is given in our co-pending U.K. application 28071/78. Although DPCM techniques will reduce the data rate to the desired levels, they can in some cases introduce a phenomenon known as "error extension". This is when a single error in transmission causes a number of errors at the receiver and occurs because the correct reconstruction of a given sample depends on the reconstructed values of previous samples.

It is an object of the present invention to mitigate error extension and this is achieved by trans¬ mitting a signal which includes both PCM and DPCM type signals. In use, the PCM type signals are used to reconstruct a DPCM sample, DPCM samples so reconstructed are preferably not used to reconstruct further samples but in any case if they are used this should only be done a limited number of times, e.g. once or twice.

In order that the present invention be more readily understood, an embodiment thereof will now be described by way of example with reference to the accompanying drawings, in which:-

Figure 1 shows diagrammatically a spatial sampling pattern; Figure 2 shows in block diagram form a transmitter according to the present invention;

Figure 3 shows in block diagram form a receiver for receiving signals from the transmitter shown in Figure 2; and Figure k shows diagrammatically a further spatial sampling pattern for use with either chrominance or luminance components signals.

The apparatus and method to be described accents a Sub-Nyquist PAL signal (2 fsc) at its input, and reconstructs a Sub-Nyquist signal after trans-

OMPI

mission.

In the composite PAL signal, colour information is carried using two amplitude modulated subcarriers of the same frequency (fsc) in phase quadrature. Therefore, when sampling at 2 fsc, each subcarrier suffers a phase reversal between samples as is shown in Figure 1. Figure 1 also shows the spatial sample pattern which results.

The basis of the method is to send one-third of the samples as p.cm. words (i.e. every third sample such as samples a,b,c,d,h, ) . Prior to transmission, these samples are used to estimate the values of the remaining samples (e,f). The differences between the estimates and the actual values of these samples is formed in the coder. The difference values may be transmitted using a lower data-rate than the original p. m. information using known techniques. At the receiver, the differential samples are reconstructed in the following way. The samples which have been received in p.cm. form (a,b,c ,d,h,g) are again used to produce estimates of the remaining samples (e,f). These estimates will be identical to the estimates made in the transmitter. The estimated value is added to the received differential value in each case to reconstruct the original information.

To send three words of the original data, it is necessary to transmit

3 x 8 = 2k bits To send 3 words of the processed information, we send: one p.cm. word = 7 bits

+ Two differential words compressed to -.-bits each = 8 bits

15 bits The transmitted data-rate is therefore reduced by the ratio 15/2-'l.

OMPI

By sending one-third of the words in p.cm. form using the particular pattern of Figure la, there are always three p.cm. samples of the correct sub- carrier phase which may be used to create very accurate estimates for each of the remaining samples e.g. the p.cm. words a,b,c would be used to estimate sample (e). Because the estimates are very accurate, the differential words may be compressed to a small number of bits without significant loss of quality. Referring now to Figure 2, 2 fsc data (at 8 bits/word) enters a series of delay elements. The delay elements consist of 1 word delays 20 and line delays 21 (566 words), and are arranged so that all of the samples a,b ,c,d,e,f,g,h are available simultaneously. The input data passes through the delay lines at a clock rate of 2 sc, and registers 22 are updated at one-third of this rate so that the appropriate samples are stored.

The equipment then produces estimates of the samples e,f using weighted sums of the pern samples :- estimate of f = .3125 g + .312 b + -375d estimate of e = .3125 a + .3125c + .375b The coefficients used may be altered.

The estimates of samples e,f are subtracted from the actual values of these samples stored in registers 23 to produce the ^' differential words ^ and A The differential words are then compressed for transmission using a ROM 2k (in Figure 2 the two words are compressed into a total of 8-bits using a non- linear coder). The differential words are multiplexed with the p.cm. words in an output register 25 (for serial transmission).

Figure 3 indicates the realisation of the decoder and is similar to Figure 2, therefore like parts will be given like reference numerals. When tho

OMPI

compressed data is received, the p.cm. words are separated from the compressed differential words. A non-linear decoder J (which has the inverse character¬ istic to that employed in the coder) is used to recreate the differential words. New estimates are made of samples e, , using the p.cm. data; these estimates should be identical to those formed in coder. The new estimates are added to the differential words to reconstruct the 2 fsc video signal. There are a number of ways in which the same algorithm could be realised. For example, if an orthagonal pattern were used (Figure l), samples from the preceding and succeeding lines could be used instead of from two lines previously and two lines later. The above embodiment applies only to composite PAL signals sampled at 2 fsc, and the essential feature is that p.c.m. data in the particular pattern of Figure 1, is mixed with differential information for trans¬ mission. In other words, a sample is reconstructed using a PCM sample of the same phase from the same line together with PCM samples of the same phase from a preceding and succeeding line.

While the above description has been given in relation to a composite PAL signal, it will be under- stood that component chrominance and luminance signals can also be treated in the same manner but separately. In this case, there is more latitude in the pattern of PCM and DPCM words transmitted because there is no need to be concerned about sub-carrier phase and it would probably be advantageous to alter the delay elements such that an immediately adjacent sample to the DPCM sample to be reconstructed were used in the production of the DPCM sample. A revised pattern of PCM and DPCM samples suitable for use with component chrominance and luminance signals is shown in Figure

This pattern is based on an orthogonal grid and is one of a number of patterns which could be used.

As will be seen from Figure 4, PCM samples are indicated by the boxes and DPCM samples are con- structed using, in the case of sample _j \, the immediately preceding sample, the same sample from the next line, and the next line, and the succeeding sample from the previous line. In the case of sample ^" b_, the DPCM sample will be constructed using the immediately succeeding sample from the same line, the. same sample from the preceding line and the preceding sample from the succeeding line. The proportions of each of the various PCM samples are Indicated on the Figure. It is thought that the apparatus required

•in order to produce such a surface as in Figure k will be obvious to one skilled in the art having regard to Figures 2 and 3«

This invention has application to terrestial lines and in digital video tape recording of T.V. signals.

OMPI