Picture processing system for television

Description Submit all comments and votes

BACKGROUND OF THE INVENTION

The invention relates to a picture processing system for television.

It is known for broadcasting studios to use television pictures derived from various sources e.g. cameras or video tape recorders. Various techniques using the source may be provided e.g. optical zooming with the camera or slow motion techniques.

OBJECTS OF THE INVENTION

An object of the invention is to provide a picture processing system which can be used in the studios remote from the picture sources which will provide electronic processing of the incoming signal which may be from a nonsynchronous source of low quality and to provide facilities for electronic reduction or expansion of picture size whilst providing an improved quality picture with reduced noise and time base corrected.

A further object of the invention is to provide a picture processing system suitable for digital line standards conversion of television signals.

SUMMARY OF THE INVENTION

According to the invention there is provided a T.V. picture processing system comprising input means for receiving T.V. picture information; input processing means for modifying the size of the picture from information received from said input means; storage means for storing the input processed picture information; co-efficient generator means connected to said store for providing modification of incoming data to said store in dependence on the co-efficient generated and on data stored in said store; output processing means for modifying the size of the picture from information received from said storage means and output means for receiving the information from said output processing means.

The input and output processors, the co-efficient generator and the store may be provided as separate elements or may be combined in an integrated store and processor system.

Means may be provided for varying the position of picture data stored in the storage means.

Thus the picture processing system of the invention may fulfil one or more of the requirements identified below:

1. The reduction in size of a television picture from full raster to a smaller raster size to give an effect of zoom down. The `compressed` picture may then be positioned and superimposed on other material generated from another picture source such as a studio announcer.

2. The increase in size of a picture from full raster size to larger than full raster size so that only a proportion of the picture normally generated from the camera or other picture source is included in the transmission. This effect gives a zoom up similar to an optical zoom system. The effect of zoom up controlled electronically without interaction with the camera gives the producer of the programme local control of his camera.

3. A requirement for zoom up in slow motion exists when an event has occurred which has been recorded on a slow motion system and then later analysis of the event is required in finer detail than was previously available from the recording made in real time. This effect requires the zoom up described briefly above.

4. Noise reduction and post production effects from high quality video tape is very valuable in making multiple generation tapes which have portions of material from various sources inserted on the final tape. There is also a requirement for zoom up and zoom down in post production techniques so that the picture material may be modified in retrospect.

5. The technique called Electronic New Gathering (ENG) which makes use of lower quality tape recorders and cameras than has been the custom in the past. The result of such quality is reduction in the broadcast images quality and a considerable increase in noise. Although improvement in light weight recorders and cameras is likely to continue the electronic system to be described makes a major contribution to this area.

BRIEF DESCRIPTION OF DRAWINGS

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

FIG. 1 shows a block diagram of the processing system of the invention.

FIG. 2 shows an embodiment for producing the system of FIG. 1.

FIG. 3 shows the area process function of the input and output processors of FIG. 2.

FIG. 4 shows the areas processed in relation to the normal T.V. picture.

FIG. 5 shows the construction of the area processor including several multipliers.

FIG. 6 shows random access memories used for the multipliers of FIG. 5.

FIG. 7 shows the volume processing function used for the input and output processors.

FIG. 8 shows the construction of a volume processor.

FIG. 9 shows the input volume processor and movement detector used in the processing system.

FIG. 10 shows an alternative volume processing function.

FIG. 11 shows an arrangement for horizontal expansion or contraction using variable write-read clock pulse frequencies.

FIG. 12 shows 12 bit data format including one bit used for a movement code.

FIG. 13 shows an arrangement for inserting the generated movement code into picture data to allow the code to be used in the adaptive output volume processor.

FIG. 14 shows the part of the processing system used for noise reduction.

FIG. 15 shows an arrangement for multilevel co-efficient settings in dependence on movement detected.

FIG. 16 shows a graph of stored co-efficients k11 which are selected in dependence of measured differences.

FIG. 17 shows a graph of co-efficients k12 stored in the ROM and accessed by the difference signal from the subtractor of FIG. 15.

FIG. 18 shows an example of a suitable movement code provided at the ROM output of FIG. 15.

FIG. 19 shows an alternative co-efficient setting system using the movement code from previous data.

FIG. 20 shows a graph of co-efficient k11 stored in the ROM and modified by the previously stored movement code as well as the difference signal produced by the arrangement of FIG. 19.

FIG. 21 shows a co-efficient setting using overall difference integration for detecting picture movement in the presence of noise.

FIG. 22 shows the area voting system for movement detection in the presence of noise or residual subcarrier.

FIG. 23 shows an alternative arrangement for effecting noise reduction with a single multiplier function.

FIG. 24 shows the requirement of a 60 to 50 field rate movement interpolator for standards conversion.

FIG. 25 shows an arrangement for effecting digital persistence in a standards converter.

FIG. 26 shows a cyclic persistence control for selecting co-efficients in dependence on field pulses.

FIG. 27 shows a full screen image with vertical sampling lines for detecting camera panning for co-efficient modification.

FIG. 28 shows an input system including a colour code generator for detecting whether data comprises luminance or chrominance components.

FIG. 29 shows the variation of co-efficient k11 with the colour code generated.

FIG. 30 shows the basic processing system including sync pulse separator and generator provided as separate units.

FIG. 31 shows an alternative system in which the input and output processors, noise reduction system and store are provided as an integrated system which can be computer controlled.

FIG. 32 shows the basic processing function using the system of FIG. 31.

FIG. 33 shows the minimum system for processing including a single processor and store.

FIG. 34 shows an arrangement for effecting the distributed store and processor of FIG. 31.

FIG. 35 shows the processor and storage system of FIG. 34 in more detail; and

FIG. 36 shows the processing function of the processor elements of FIG. 35.

DESCRIPTION OF PREFERRED EMBODIMENTS

The processing system of FIG. 1 shows an input receiving unit 1 which receives an incoming T.V. signal (e.g. NTSC 525 line standard). This input unit 1 will modify the incoming signal as necessary to a format acceptable to the compression/expansion unit 2. Thus if the signal is normal composite video, the chrominance and luminance components will be separated and the analogue information converted into digital form for application to the compression/expansion unit 2. Such analogue to digital conversion of T.V. signals is well known. For compression, the unit 2 will take the digital data corresponding to incoming picture point information and derive a reduced number of picture points for a given picture size so that the reduced size picture from the system output will be produced in real time derived from the incoming data. For expansion, unit 2 will modify the incoming picture information so as to derive a larger number of picture points than originally present for a given picture size so that at the system output an expanded picture based on the original data will be produced in real time.

The modified data from unit 2 will be temporarily stored in digital store 3 prior to read out (e.g. for 1 frame period). A compression/expansion unit 4 may also be provided at the output of store 3. Thus the unit 2 could conveniently be used for compression and unit 4 for expansion. The amount of compression or expansion can be variable if required so that the zoom down or zoom up can be provided in real time. The modified picture data is received by output receiving unit 5 which will convert the data into analogue form and effect the known output processing functions to provide an analogue composite T.V. signal at output 55. The data in store 3 is accessible to allow modification by coefficient generator unit 6. The generator unit 6 produces coefficients to modify the stored data so as to provide an enhanced picture at the system output.

In the general case where the compression/expansion is variable the coefficient generation unit 6 will effect noise reduction on the data. If the amount of compression or expansion is fixed as in the case of standards conversion i.e. 625 to 525 lines (compression) or 525 to 625 lines (expansion) then the coefficients generator is used to provide movement interpolation coefficients which is explained in greater detail below.

In addition to compression and expansion, a position control unit 7 may be provided to vary the relative position of the picture on the normal T.V. frame so that for example a reduced picture may be moved from a central position to one corner of the screen.

An embodiment for producing the desired functions of the system of FIG. 1 is shown in block diagram form in FIG. 2. The system of FIG. 2 is described for producing variable compression and expansion and the coefficient generator 6 connected to the store is used to provide noise reduction coefficients. Compression/expansion unit 2 is used for compression and unit 4 is used specifically for expansion. It will be understood that these functions can be interchanged.

The FIG. 2 arrangement is described for use with NTSC line standard but could be adapted to other standards.

The input composite television signal comprising chrominance and luminance information is applied to input 10 of input receiving unit 1 having a decoder 11 therein which produces separate signals for luminance and the colour difference signals. Luminance is designated Y whilst the two colour difference signals are categorised as I and Q. The luminance signal has a bandwidth of 4.2 MHz whilst the colour difference signals I and Q each have a bandwidth less than 1 MHz.

The Y, I Q signals are applied to an analogue multiplexer 12 which looks at each of them on a time shared basis such that the sampling frequency is sufficient to convey all of the information. A typical sampling frequency for Y will be 10.7 MHz and for I and Q 3.58 MHz. The time sharing may be arranged in any sequence so that the sampling frequency is maintained above the minimum required by information theory. The minimum frequency which may be used in this system is twice the maximum bandwidth required at the output of the system for Y, I and Q.

The analogue multiplexer produces a time shared serial analogue data stream which is applied to a sample-hold unit 13 which stores the information presented sufficiently long for an analogue to digital conversion to take place in ADC 14.

The analogue to digital conversion produces a digital data stream M bits wide. In the system described herein M lies between 8 and 10 bits. 8 bits is sufficient to convey all analogue input information without significant signal degradation. The output from the ADC of input unit 1 is applied to the compression unit 2.

The 8 bit wide digital data is applied to a process input formatter 15 within compression unit 2 which formatter accepts the digital data stream in the order presented, stores it and represents it so that the input area processor 16 may operate on the signal.

The input processor 16 is an area processing operation which accepts a number of picture points from adjacent lines horizontally and adjacent points vertically. A co-efficient (described in detail below) is applied to each of the input points and the resultant output is a single data word for each new picture point which is the sum of various proportions of the input data points over the area being processed. The required compression coefficients are stored in a coefficient store 26 and the degree of compression can be controlled by compression control 28 via ADC 27.

The input processor buffer store 17 accepts data from the input processor 16 at the rate presented and re-formats it for subsequent storage in the main store 3 via the noise reduction system 6. The noise reduction input co-efficient modifier 18 of system 6 accepts the digital data stream and modifies it by a factor designated K11. The output from co-efficient modifier 18 is applied to one input of an M bit wide adder 19. The output of the adder 19 is Q bits wide and applied to co-efficient modifier 20 which accepts the input from the adder, modifies it by a coefficient K13 and provides the output Q bits wide available for the input port 1 of data store 22 within store unit 3.

Port 2 of data store 22 is arranged to read data from the data store and apply it to co-efficient modifier 23. Modifier 23 accepts the information from the data store, modifies it by co-efficient K12 and applies it to the other side of the Q bits wide adder 19.

The co-efficients K11, K12 and K13 are set by the set co-efficients unit 25 which examines data from the input process buffer store 17 and from the output port 2 of data store 22 in an area comparison system. The result of the area comparison information modifies the co-efficients on a point by point basis in real time. The noise reduction data is stored in noise reduction store 41. Noise reduction can be controlled by control 43 via ADC 42.

Port 3 from the digital data store 22 produces data R bits wide and applies it to the output processor buffer formatter 30 of expansion unit 4. The formatter 30 accepts the output from port 3 and modifies it for processing in the output processor unit 31. This modification is a simple re-arrangement of the data necessary for use in the output processor.

The output processor 31 operates as an area processing function in a similar way to the input processor. The output processor accepts data from a number of adjacent lines horizontally and a number of adjacent picture points vertically. Each of the picture points are modified by a co-efficient described below and the resultant output data R bits wide is available for application to the output processor buffer store 32.

The co-efficients for expansion are stored in store 45, and the degree of expansion can be controlled by control 47 via ADC 46. Enhancement can also be provided by control 49 via ADC 48 and is described in more detail below.

The output process buffer store 32 takes the information from the output processor function and modifies it for application to the digital to analogue converter 36 of output receiving unit 5. The modification is a simple re-arrangement and re-timing of the output information such that it appears in the time scale necessary for handling by the digital to analogue converter(DAC).

The DAC 36 accepts data R bits wide and produces an analogue output which is a true representation of the digital number presented to the input.

The analogue output is applied to three sample-hold units 37, 38, 39, one of which is each allocated for Y, I and Q. The resultant analogue outputs Y, I and Q are then applied to an encoder 40 to reproduce a composite NTSC television signal output.

In the system under consideration the resolution of the data store is Q bits wide where Q lies between 10 and 12.

The output data from port 3 of the data store 22 is R bits wide where R lies between 8 and 10. The system is capable of operating without any signal degradation at 8 bits wide and signal enhancement may be utilised so that 10 bit data is available to provide a significant noise reduction.

The system of FIG. 2 will now be described more fully. The input de-coder 11 is a standard television equipment which accepts a composite T.V. waveform, filters it to exclude the chrominance information carried on the sub-carrier of 3.58 MHz and coherently detects the I and Q components. The YIQ output is filtered to provide full bandwidth signals on each of the three channels.

The analogue input multiplexer 12 is a simple known analogue switch operating at high speed. A bridge diode switch has proved one way of producing an analogue multiplexer capable of operating at the speeds required which are in the region of 15 MHz between switch points.

The analogue sample-hold unit 13 before the ADC comprises a high speed bridge diode switch and memory capacitor which retains the stored charge sufficiently long (66 nanoseconds)for the analogue to digital conversion to take place.

The analogue to digital converter 14 is of known construction and can be of a form described in British patent application 26613/74 (U.S. Pat. No. 4,005,410). The basic analogue to digital conversion produces parallel digital data 8 bits wide for application to the input process buffer formatter. The sequence of data from the ADC corresponds to the time shared operation of the analogue multiplexer and may for example be in the form YYI, YYQ. In the form described the input area processor requires a format YYY, YYY, III, YYY, YYY, QQQ. The processor input buffer formatter 15 accepts the data as presented by the ADC and reformats it for application to the input processor. It is simply a buffer store operating at approximately 15 MHz (e.g. 1 line stores of 1024 locations).

The input processor 16 operates in the area processing mode. FIG. 3 shows the function of the input area processor. Successive picture points P1 to P9 on adjacent lines N, N+1 and N+2 are applied to the input processor. Each of the points P1 through P9 is modified by co-efficient K1 through K9. The resultant sum is a new picture point designated NP1 where

NP1=K1P1+K2P2+K3P3 . . . etc. through K9P9.

if the input area processor function is operating to reduce the picture size for compressed pictures the output data may appear more slowly than the input data. Thus for compression, the number of new picture points produced will be less than the original number of picture points but each new picture point will be derived from data on the nearest 9 picture points. How the processor effects this function will now be described.

In order to reduce the size of a standard television picture interpolation is needed across the picture area. In the system described the total television picture is broken up into a number of picture points. The picture could typically be broken into 512 picture points per line for a 525 line picture. As already described with reference to FIG. 3, the area process function is effected for new picture point NP1 by the expression

NP1=K1 P1+K2 P2+. . . K9 P9.

this area is designated Area A in FIG. 4. When calculating the next picture point NP2 (say) the values of coefficients K1 to K9 for Area B will be different to those for Area A thus

NP1=K1A P1+K2A P2+K3A P3+ . . . K9A P9 and

NP2=K1B P2+K2B P3+K3B P10+ . . . K9B P12.

thus the input area process remains the same but the co-efficients K1 through K9 are variable.

The operation of area interpolation occurs in real time and as the data represents incoming information scanned horizontally the co-efficients K1 through K9 have to change across the length of 1 television line. In the system described the switch occurs between picture points.

In the same way vertically the boundaries between the lines represent co-efficient changes. Each new picture point is computed from information available from the nearest 9 picture points to that new picture point.

In order to switch co-efficients between the picture point boundaries horizontally excess look-up tables are provided within the basic system. However as it is possible to re-load data into the look-up tables when they are not in use it is possible to implement the system utilising only one complete set of excess look-up tables.

The co-efficients K1 through K9 are stored in a separate co-efficient storage unit 26. The required degree of compression is manually controlled by the analogue type control 28. The amount of compression is converted into a digital number in analogue to digital converter 27 and applied to the co-efficient store so that the required values of K1 through K9 are extracted for each setting of the compression control.

The area processor 16 is shown in detail in FIG. 5. Multipliers 60-68 each receive data on one picture point (P1-P9) and multiply the data by co-efficients K1 to K9 respectively, which co-efficients will each be variable but preset. The modified data is added in adder 69 which comprises a 9 input .times.8 bit adder. The output from adder 69 will be the new picture point NP1.

The co-efficient multiplier function of area processor 16 (i.e. multipliers 60-68) can be effected by using random access memories (RAM), see FIG. 6. The RAM 70 shown is of 8.times.256 bit capacity and such memories and their mode of operation are well known in digital processing. The coefficients K1 to K9 are loaded into the store locations within the RAM during a write cycle. The co-efficient data from the co-efficient store 26 (of FIG. 2) is applied to the RAM data input 71 shown in FIG. 6. The location to which data is written in is determined by store address data input 72. Address data is applied in the normal way to the address input 72 to input the co-efficient data at input 71. The addressing data is shown as `load co-efficients`. During operation as a multiplier (i.e. read cycle) the incoming video data is applied to the RAM `address` terminals 72. The RAM has sufficient addresses so that each input number identifies one particular location within the store. Thus as each location has a preloaded co-efficient stored therein when a particular location is accessed (i.e. in dependance on the incoming data which effectively defines the address), the data stored in a particular location is read out from the RAM at output 73. This data will either be an 0 or I depending on the predetermined co-efficient. Thus the 8 bit input data for picture point P1 will effectively be multiplied by a coefficient K1.

The input processor buffer store 17 accepts the data produced by the input area processor 16 and stores it ready for input to the data store. It is simply a buffer store operating at a maximum of 15 MHz and a figure which may be lower in the case of compressed pictures.

Co-efficient modifier 18 includes a multiplier operating in real time. In this system there are a number of multipliers and a basic requirement is the ability to multiply at high speed. The method used is a look-up table and is applicable in all the co-efficients (see explanation above in relation to the multipliers of area processor 16). The input data from buffer 17 is applied to a RAM within modifier 18 which has sufficient addresses so that each input number identifies one particular location in the store. As explained above the video data is applied to the terminals usually known as the `address` terminals. At the location identified by the data either a 0 or a 1 is stored and read onto the data output. The co-efficient K11 is pre-determined and pre-stored as a series of 0's and I's in the locations within the RAM (e.g. 8.times.256 bit).

As for the RAM of the processor 16, in order to load the co-efficients the co-efficient data is applied to the terminals marked co-efficient data input and the addresses are multiplexed to the load co-efficients.

Once the co-efficients have been loaded the address terminals are connected to the data input and the RAM store is operated in the read mode.

The output from co-efficient modifier 18 is applied to one side of an M bit.times.Q bit wide adder. Standard arithemetic elements are used for the adder 19.

Co-efficient modifier 20 takes the form of a look-up system as described above, the output being applied to the data store.

Co-efficient modifer 23 also takes the form of a look-up system.

Co-efficients K11, K12 and K13 are set in the set co-efficients unit 25. The operation of this block is basically to look at the output data over a small area and compare it with the new input data which is appropriate for the same area. The co-efficients are modified depending upon the amount of difference which exists between the data. Preset co-efficients K11, K12 and K13 determine the amount of varying degrees of noise reduction which may be applied to the system. The output from analogue noise reduction control 43 is applied to analogue to digital converter 42 which is connected to the noise reduction data store 41. The general principle of noise reduction is in the form of a re-circulating digital number to which a proportion of the new input information is added and a proportion of the total removed at each store location within store 22. The system may be likened to an integrator with leakage. Picture information in the television system contains a large amount of stationary data during which high noise reduction co-efficients may be applied. The nature of the noise may be random and the larger the integration time in the data store, the greater the reduction in noise.

Co-efficient K12 controls the amount of output data fed back and re-stored.

Co-efficient K13 controls the amount of data which is removed during each store cycle. Co-efficient K11 is provided to prevent the system overflowing and exceeding the store capacity.

The amount of noise reduction effected as explained above is dependent on the co-efficients K11, 12, 13. The noise reduction control in principle alters the amount of integration applied. For static picture information (e.g. T.V. test card) it is readily apparent that the amount of picture feedback (as determined by co-efficient K12) can be high since the next frame will be identical to the previous frame. Thus values for the co-efficients during static information may be K11=0.1; K12=0.9; K13=0.95.

In the case when the T.V. picture is not static, (e.g. when scene movement is occuring) to prevent distortion to the noise reduced picture it is necessary to have a shorter integration period; the faster the movement the shorter the allowable integration time. Thus the values of co-efficients K11, K12 and K13 will have to be adjusted accordingly. Typically for high movement the co-efficient values could be

K11=1.0; K12=0; K13=1.0.

the data store 22 has three data ports 1, 2 and 3. Port 1 allows data to be written into the store, port 2 allows data to be read from the store at a location corresponding to port 1 and port 3 allows data to be read from the store at another location. In principle the three ports run asynchronously. The data store may take the form of a system described in British patent application No. 6585/76 (U.S. patent application Ser. No. 764,148). The store is large enough to store at least one complete television frame of Y, I and Q information at full bandwidth (5-6 M Bits).

The store 3 has an associated store control which includes address counters for addressing the various memory addresses within the frame store for a read or write cycle and timing control for producing timing signals for addressing the memory elements at the correct point in time, in known manner and as explained in detail in the above referenced Patent Application. The store itself is constructed from known 64.times.64 bit memory chips (i.e. 4096 by 1 bit RAM) which bit locations are accessed by entering row and column address information from the store control as an 18 bit address derived from picture point counters. As the 4096 random access memory chip is dynamic, a refresh cycle must also be effected to retain the stored data. Refresh address counters are therefore included. For example, a store using 16 cards each containing 32 RAM chips would provide a framestore of 256.times.512 words 8 bits wide to allow 512 video lines each of 512 picture points to be stored.

The storage capacity could be expanded as required.

Port 3 provides digital data to the output processor buffer formatter 30. The output area processor 31 is similar in concept to the input processor but operates with R bits wide data instead of M bits wide data. R is equal to or greater than M for picture noise reduction.

The limit to the noise reduction which is not limited by systematic errors is dependent on the capacity of the data store. In the system under discussion, with suitable storage, 16 complete frames of information may be stored and integrated before the store overflows using:

M=8

Q=12

At this level of integration a reasonable value for R is:

R=10

The output area processor 31 is similar in concept to the input area processor accepting inputs from adjacent lines and adjacent points. Co-efficients are applied using the look-up system and the resultant output information is stored in the output processor buffer store 32.

The output processing system allows the image to be enlarged. Information from the part of the store being accessed is read into the output process buffer 30 and processed using co-efficients for K1 through K9 stored in the co-efficient storage for expansion unit 45. The amount of expansion required is controlled by the expansion control 47. The analogue to digital convertor 46 enables the values for K1 through K9 on the output area processor to be withdrawn from the co-efficient store 45. The implementation of the output area processor is similar in every respect to the input area processor. In principle the output area processor may be used for compressed pictures as well as enlarged pictures. The only difference is in the values of the co-efficients required.

In addition to the basic expansion function, picture enhancement can be effected. Picture enhancement is generally concerned with improvements in edge effects. In the television system this is called horizontal and vertical aperture correction. The output area processor enables both horizontal and vertical aperture correction to be undertaken by selecting appropriate co-efficients for K1 through K9. Enhancement control 49 is connected to the store 45 via ADC 48 and operates in similar manner to the expansion control.

The area processed digital data is passed to the output process buffer store 32 which buffer applies this digital data to the digital to analogue converter 36 which converts the digital number to an analogue representation. The DAC may be of the form described in British patent application 25721/73 (U.S. Pat. No. 472,059).

The three sample-hold units 37, 38, 39 are used to store the values of Y, I and Q in analogue form. The basic sample-hold unit is a diode switch and memory capacitor.

The encoder 40 is a standard piece of television equipment which accepts sync pulse inputs, sub-carrier inputs and Y, I and Q values. The inputs are combined to produce a standard composite NTSC T.V. signal output at the system output 55.

In the system which has been described 3 horizontal lines and 3 adjacent vertical points are processed as an area. There is no reason why a larger number of points cannot be used for very large magnifications and very large enhancements.

Similarly if only a small range of compression, enlargement and enhancement is required less than 9 points may be processed as an area.

There are clearly a number of points which will be considered in optimising the values of the co-efficients used in this system. One point which is worthy of mention is the ability of the system to be given a designated frequency response at sub-carrier. De-coding of television information into Y, I and Q does not always completely remove sub-carrier information. Residual sub-carrier in this system may be removed by utilising particular values of the co-efficients.

Difficulties in producing a line by line de-coder which does not suffer from degradation is a limiting factor in the overall performance of the equipment. In this equipment a complete framestore is available and may be utilised to assist the decoding operation. The format of NTSC television signals is such that the phase of the sub-carrier is exactly 180.degree. out on a frame by frame basis using the same reference picture point by adding together two successive frames the sub-carrier information may be reduced to zero. Utilisation of this principle in the de-coder assists in maintaining full luminance resolution horizontally and vertically.

By incorporating in the system a data store 22 which has totally asynchronous read and write ports, it is clear that this enables the whole equipment to be operated in environments with asynchronous television inputs. For example the equipment may be located in the studio and may be used to operate on a remote source which in no way is synchronized to the studio. The system then becomes a fully synchronizing T.V. picture processing system (see also British patent application No. 3731/76 and U.S. application Ser. No. 7,641,317).

The discussion of data compression and expansion has not so far covered which portion of the picture is to be used for display. The allocation of addresses in the data store for read or write operations may be offset by the position control unit 7. Information from the horizontal and vertical position controls 51, 53 allows information to be withdrawn from the position storage unit 50 via ADC 52, 54 and applied to the addresses being used in the main data store 22. Thus the store address counters within overall store unit 3 can be incremented/decremented so that picture data can effectively be shifted up/down and/or right left relative to the normal picture position. The compressed picture may be inserted at any part of the raster using the horizontal and vertical position controls. Similarly any part of the expanded picture may be examined using the same control when operating in the expanded picture mode.

The equipment has been described so far in terms of picture manipulation for pure compression about a central point and expansion about a central point. It is clear however that compression about any point or axis in the system may be undertaken by altering the co-efficients in the input area processor and output area processor. If for example the co-efficients are calculated in the normal way across the picture that is to say, co-efficients for the first picture point are different from those for the second picture point and different from the third, etc but that vertically all co-efficients remain the same then compression occurs about the centre line of the picture rather than the central point. This effect is called `horizontal squeezing`.

With regard to the overall picture processing arrangements, intrinsic in the system is the capability of operating with video tape input signals. The design of the inherent store timing arrangements described in the aforementioned patent specifications are such that digital time base correction takes place on the incoming signal. The digital time base correction is the subject of other co-pending patents and will not be described in any detail. In this particular equipment the main requirement is the ability to utilise poor quality signals which generally record using a colour processing system known as `heterodyne`. The output from such a system gives a stable chrominance signal with a time varying luminance signal. The input of this equipment as explained can accept such a signal.

As already mentioned the system described above may be used as a digital standards converter. Standards conversion generally involves re-arrangement of the television picture so that the outgoing television standard has a different field and line frequency to the incoming standard. For example, PAL to NTSC standards conversion will require taking an incoming line standard of 625 lines per frame and converting this to an outgoing standard of 525 lines per frame. Thus the number of lines is reduced and is comparable with a fixed amount of compression.

In addition to the reduced number of lines the number of fields per second will change. For European PAL the field rate is 50 fields/sec and NTSC used in USA is 60 fields/sec. The reduction in the number of lines is effected by the input area processor (area interpolation) and the increase in field rate can be accommodated due to the asynchronous nature of the frame store which allows different write in and read out rates. A proportion of the data from the old frame is combined with a proportion of the new frame using the `leaking` integrator system i.e. co-efficients K11, 12, 13 as described above to produce smooth movement (movement interpolation) at different frame rates.

In addition, normal picture compression, enlargement and enhancement may also be effected.

The system described in FIG. 2 requires a storage of a large number of co-efficients. There are 9 variable co-efficients on the input processor, 9 variable co-efficients on the output processor and 3 variable co-efficients in the noise reduction mechanism. Each of the processing co-efficients is different for the various picture points and further each co-efficient is different for each size of compression or expansion. Noise and movement in the picture give rise to the need for variable co-efficients in the noise reduction system.

As already discussed, the co-efficients K11, K12 & K13 have to be varied to take into account picture movement so as to avoid distortion. It is possible to detect such movement so as to vary co-efficient selection accordingly. This is effected by comparing data changes on the picture point information. For example, if each picture point in an incoming picture is subtracted from the data previously stored for the corresponding location for an earlier picture and if the difference signal exceeds a threshold level (e.g. using a comparator) then the picture is deemed to have moved and co-efficients K11, K12 and K13 are switched to levels which would be appropriate for movement. Whilst any detected change remains within the threshold level, co-efficients suitable for noise reduction of still images are retained.

As a refinement to improve the system flexibility, one or more suitably programmed digital microprocessors could be used to calculate the required co-efficients.

In addition, they can be used to determine the address locations fixed in store and to calculate the interaction of the controls for compression, expansion, picture enhancement, noise reduction, horizontal position and vertical position with the store locations and hardware co-efficient look-up tables.

In the system described with relation to FIG. 2, the generation of a new picture point is effected by taking into account information from the surrounding picture points (i.e. area manipulation).

The basic requirement is the synthesis of a picture point which did not exist as a picture point on the incoming video data. The engineer aims at producing the best estimate of the likely value of a picture point by examining picture points around the synthesised picture point and either adding or subtracting various proportions of them to produce the best result.

Theoretical studies give a good guide to the value