Apparatus and Method for Processing Television Picture Signals and Other Information

This invention relates to information processing apparatus comprising a memory means for writing information into the memory at times depending on first timing signals, and means for reading information out of the memory at times depending on second timing signals which drift in and out of phase with the first timing signals. The invention was conceived when designing a picture synchronization system for television signals.

Many forms of television picture processing depend for their operation upon the storage of one or more fields of the signal. Examples of types of picture processing which employ storage are: standards conversion; picture synchronization; noise reduction; and picture size reduction and expansion.

Picture storage can be carried out using analog methods, the use of quartz delay lines being an example of this type of storage. However, digital memories are cheaper, more reliable and more compact and most equipment employing picture storage now uses digital semiconductor memory chips with the television signal being converted from analog-to-digital form at the input of the equipment and back to analog at the output.

This invention is particularly, although not exclusively, applicable for use in a picture synchronizer. The purpose of a synchronizer is to re-time a television signal, e.g. one arriving from an outside broadcast, so that its subcarrier, line and field frequencies are identical to some local reference, e.g. the studio camera pictures. At the output of the synchronizer the remote signal can be treated as if it were generated locally and can be mixed, faded, and inserted into local pictures without causing timing disturbance to the transmitted signal.

A typical known synchronizer works by writing the incoming signal into the memory at a rate determined by the incoming signal and then reading the signal out of the memory at the rate fixed by a local reference signal. Since there is no frequency or phase relationship between the reading and writing operations, there will be occasions when the reading will overtake the writing or vice-versa. Also there will be occasions when reading and writing operations coincide.

It is a feature of most random access memory systems that, at any instant, only one memory location may be addressed, and that information may be written into or read from the chosen location, but not both together. One known method of overcoming the need for simultaneous writing and reading within a memory chip is to provide storage additional to that strictly needed for one field of the picture. In this way, when an attempt is made to read from and write into the memory at the same time; one of the operations, e.g. writing, is transferred to an additional memory. This known method suffers from the disadvantage that substantial additional memory is required. One object of this invention is to avoid this disadvantage.

This invention provides a facility for delaying at least some of the writing times so that they do not coincide with the reading times. Thus, the read and write operations are `interleaved`, i.e. the memory is addressed in a read-write-read etc. sequence and no additional memory is required.

The writing times are preferably delayed by varying amounts, chosen so that each writing time is approximately mid-way between reading times.

Since it may be necessary to write information into and read information from the memory at a very fast rate it is preferable to include a serial-to-parallel converter at the input to the memory and a parallel-to-serial converter at the output. This allows the memory sufficient time to register each bit.

When the invention is used in a synchronizer for television signals, the said `first` i.e. "write" timing signals are preferably generated from the line synchronizing pulses, the field synchronizing pulses and the burst of the signals to be processed, whereas the `second` timing signals are generated similarly from an independent source. In this particular application of the invention the frequencies of the first and second timing signals are marginally different. This means that they drift into and out of phase with each other. When they are in phase the delay produced in accordance with the invention avoids the memory being instructed to read and write simultaneously. When they are 180.degree. out of phase there is no need for delay.

The change in phase of the `first` and `second` timing signals in a television synchronizer is very small during any given line. It is, therefore, satisfactory to set a delay time which remains constant during any given line of the signal to be processed.

The blanking periods, between lines of the video information being read out of the memory are not normally equal to a multiple of the period between reading times. This means that the delay, fixed for a given line of the signal to be processed is inappropriate after the blanking period of the signal being read out of the memory. This problem can be overcome by delaying the writing times by a further fixed amount after the blanking period.

In some other applications of the invention the first and second timing signals can be the same in which case the delay will be fixed value, permanently present. Such would be the case for example in picture size reduction systems and noise reduction systems for television signals.

FIG. 1 is a block diagram showing, in full lines, a television picture synchronization device designed to synchronize television signals derived from somewhere outside a studio with signals generated locally, i.e. in the studio; and, in broken lines, a modification applicable when the memory requires more than one write clock.

FIG. 2 shows waveforms A, B, C, D, E, F, G and H which are generated at various positions, also indicated on FIG. 1 by the same reference letters; and

FIG. 3 shows waveform A of FIG. 2 but on a smaller time scale, and also waveforms E and J on the same (smaller) time scale.

The circuit of FIG. 1 is designed for use in a television studio and the object is to synchronize a video television signal derived from a remote source with timing signals produced locally in the studio.

The remote signal is received by an input 1 and is digitized by an analog-to-digital converter 2 into 8-bit parallel form, samples being taken at a frequency of three times the frequency of the subcarrier (3 fsc). This video input 1 also drives a timing generator 3 that produces first timing signals constituting line frequency pulses A, 3 fsc clocks B, and write clock pulses C. The write clock pulses C are started by line pulses A and have a period of repetition which is a predetermined number of periods of the 3 fsc clocks B. In the described embodiment this is fifteen times the period of the 3 fsc clocks. Local pulses generated within the studio pass through a second timing generator 4, similar to generator 3. The generator 4 produces second timing signals constituting line pulse E and read clock pulses D. These are similar to A and C respectively, but are not locked to the remote signal.

A comparator 5 compares the phase of the remote line pulses A relative to local read clock pulses D and produces a signal F which opens a NAND gate 6 to pass 3 fsc clock pulses B during a period between the beginning of a pulse A and the beginning of the next pulse D. The number of clock pulses B passed by the gate 6 during this period is a measure of the relative timing of pulses A and D. A divide-by-fifteen counter 7 counts the clock pulses G passed by the gate 6 and the number contained in the counter when the clocks stop is stored in a latch 8 for the duration of a television line. The counter 7 is cleared on each line just before the gate 6 is enabled. FIG. 2 shows a situation where the time difference between the beginning of a pulse A and the beginning of the next pulse D is such that nine clock pulses G are passed to the counter 7. The output of the counter 7, which is in four-bit parallel form, is held for one line period in the latch circuit 8, and the output of the latch circuit is fed to two identical selectors 9A and 9B.

The selector 9A feeds each write clock pulses C to a selected one of fifteen output connections 10A. These connections 10A are connected to respective inputs of a fourteen-bit shift register 11A so as to produce pulses H which are similar to pulses C, but are delayed in accordance with the following table:

When a count of eight is held by the latch circuit 8 the selector 9A feeds the pulses C to one particular output which is connected directly to an OR gate 12A. This means that the pulse H is produced simultaneously with a pulse C; i.e. there is no delay.

The effect of the delays caused by the circuits 7, 8, 9A, 10A, 11A and 12A is to delay the write pulses C by a time such that the leading edges of the delayed write pulses H are approximately half-way between leading edges of the read clock pulse D. The delay is reset for each line of the remote video signals but, except for a fixed adjustment produced by circuits 13A, 14A and 15A to be described later, remains constant during each line.

The delayed write pulses from the gate 14A are fed to the "write clock" input of a random access memory circuit 16. The circuits 28 and 29, shown in broken lines, are not included in this particular embodiment of the invention. They will be referred to later in relation to a modification. Each write pulse enables the memory to receive video information from a serial-to-parallel converter 17.

The video information from one output of the analog-to-digital converter 2 is delayed in exactly the same way as the "write pulses", but by separate circuits 10B, 11B, 12B, 13B, 14B and 15B. These are identical to circuits 10A, 11A, 12A, 13A, 14A and 15A.

The delayed signals, representing one output of the analog-to-digital converter 2 cannot be fed directly to the memory 16 since they occur at a frequency of 3 fsc which is too fast to register in the memory. They are, therefore, converted by circuit 17 into 15-bit parallel form before being entered in the memory 16.

The remote video information stored in the memory 16 is read out by the read clock signal D derived from the local video signal and, therefore, the output from memory 16 is synchronized with the local video signal. Because of the delay introduced by the circuitry, the memory is never required to read and write simultaneously.

The output from the memory is passed to a parallel-to-serial converter 18 and thence to a digital-to-analog converter 19. The latter also receives, through inputs 20, the outputs from seven other circuits, identical to that illustrated, which process signals from the seven other outputs 21 of the analog-to-digital converter 2.

The output of the digital-to-analog converter 19 represents the remote picture signal synchronized with the locally generated signal.

FIG. 3 shows at A and E respectively the blanking periods of a line of the remote video signal (before the delay) and of a line of the local video signal. It should be noted that the blanking period shown at E of the local video signal is exactly the same as the blanking period of the remote signal withdrawn after processing from the memory 16. The picture signals of the remote signal occupy a part 22 of the line during which 765 samples are taken from the analog-to-digital converter 2. The blanking period 23 occupies 86.25 sample periods. The signal E is similar, but its blanking period may, and normally will, occur during the picture part of the remote signal. Since the delay to the individual samples of the remote signal is set by the latch circuit 8 for a whole line period, this will be correct only for a time interval indicated at 24. It will be wrong during the time interval 25. This can be corrected by introducing an additional delay during the interval 25. Since the read and write clocks occur every 15 samples, the correct delay is equal to the remainder of 86.25/15 sample periods; i.e. 11.25 sample periods. Thus, an additional delay of this amount is introduced to the remote video signals during the interval 25. This is done by the circuits 26, 13A, 13B, 14A, 14B, 15A and 15B shown in FIG. 1. The comparator 26 receives the remote line pulses A from the timing generator 3 and the local line pulses E from the timing generator 4 to produce an output as shown at J on FIG. 3. When the waveform J is at the low level it sets the switches 13A and 13B in conditions where they connect the outputs of gates 12A and 12B directly to the OR gates 14A and 14B respectively. When the waveform J is at the high level, the switches 13A and 13B are set in conditions such that the outputs of gates 12A and 12B are connected to delay circuits in the form of 11-bit serial registers 15A and 15B. The outputs of these registers are connected to the OR gates 14A and 14B. Thus, during the appropriate interval an additional delay of 11 periods of the waveform B is introduced. This is close enough to the required delay of 11.5 periods to ensure that the read signals at the input to the memory are approximately mid-way between the write signals.

If the memory 16 requires more than one write clock, all three must be delayed by the same amount. A simpler solution is to modify the circuit of FIG. 2 so as to delay the 3 fsc clocks B, instead of the write clocks C. Delayed write clocks can then be generated from the delayed 3 fsc clocks. The necessary additions to the circuit are indicated in broken lines in FIG. 1. Also, the lowermost output of the timing generator 3, as shown in FIG. 1, must be disconnected.

In this modified circuit the input to the selector 9A is connected, through a divide-by-two circuit 27 to receive the 3 fsc pulses B instead of clock pulses C. The divide-by-two circuit 27 is needed simply to enable identical delay circuits to be used to delay the video signals and the clock signals. A multiply-by-two circuit 28 is included to bring the frequency of the delayed signals back to 3 fsc. The output of the circuit 28 drives a clock generator 29 which produces as many clock signals as may be required and feeds them to the memory 16.