Multiple Mode Controlling Signal Processing Apparatus

1. Field of the Invention

This invention relates to controlling signal processing.

2. Description of the Prior Art

It is common for signal processing apparatus to be operable in a number of signal processing modes. An example of such an apparatus is a video down-converter which can receive an input video signal in, say, an 1125 line or a 1250 line high definition video format and generate an output video signal in, say, a 525 or a 625 line conventional definition format. The various possible modes of operation, in addition to the simple choice of line format, might include a selection between a horizontal line cropping mode and a horizontal spatial compression mode to cope with the different aspect ratios of the input and output signals, and a selection between a full screen and a "letterbox" mode for the output video signal.

Accordingly, an apparatus of this type can have many different modes of operation, possibly with a normal or default mode applying if no other mode is positively selected.

In the case of a signal processing apparatus of the type described above, it is then necessary to ensure that the appropriate signal processing coefficients (e.g. filter coefficients and various other signal processing parameters) are passed to signal processing devices within the apparatus.

This invention provides signal processing apparatus comprising: a control read only memory storing:

(i) two or more sets of signal processing parameter data to control signal processing operations of the apparatus; and

(ii) configuration data specifying one of the sets of signal processing parameter data stored in the control read only memory;

a control read/write memory capable of storing:

(i) at least one set of signal processing parameter data; and

(ii) configuration data capable of specifying a set of signal processing parameter data stored in the control read only memory or a set of signal processing parameter data stored in the control read/write memory; and control logic, responsive to an input selection signal specifying whether the configuration data in the control read only memory or the configuration data in the read/write memory should be selected for use, for controlling the signal processing apparatus to perform signal processing operations in accordance with the signal processing parameter data specified by the selected configuration data.

The invention therefore provides an advantageously convenient way of allowing three possible modes of operation of the apparatus:

(i) if a default mode of operation is required, it is not necessary to load anything into the control read/write memory. The signal processing parameter data to be used is specified by the configuration data permanently written into the control read only memory, and the required parameter data is found in the control read only memory.

(ii) If, however, a different mode of operation is required but using alternative parameter data still pre-programmed in the control read only memory, it is merely necessary to write a different version of the configuration data into the control read/write memory. This will then provide pointers to non-default sets of parameter data stored in the control read only memory.

(iii) In a further possibility, if a mode of operation is required which cannot use the possible sets of parameter data stored in the control read only memory, replacement parameter data can be stored in the read/write memory and pointed to by the configuration data held in the read/write memory.

Another feature of video down-conversion apparatus of the type described earlier is that, for high quality video output, horizontal and vertical interpolation digital filters having a large number of taps must be used. A correspondingly large number of filter coefficients (e.g 11 or 13) are therefore required for each filtering operation.

However, not only does each filtering operation require a large number of coefficients, but also many such sets of coefficients may be needed during generation of, say, an output video field. This is a particular problem for a vertical interpolator. In general, an output video line will not coincide spatially with the vertical position of an input line. Depending on the relative position of a current output line with respect to the input lines, a different set of coefficients will be required for the filtering process performed by the vertical interpolator. For example, if n possible sets of coefficients are provided, the vertical position of an output line can be specified to an accuracy of 1/n of the input line spacing.

Typically, the number of possible sets of vertical filter coefficients to achieve high quality down conversion might be 32. It is likely that all of these sets will be required at different stages during the generation of a single output field, as the relative vertical position of each output video line with respect to the input lies varies down the field. Accordingly, in this example, in one field the number of coefficients required will be

13 (taps).times.32 (coefficients per set) and the storage capacity required at the vertical interpolator will be, assuming that each coefficient is specified to a ten bit accuracy, 13.times.32.times.10 bits=4160 bits

In an apparatus fabricated using discrete components, the need to store this amount of data at the vertical interpolator is not so much of a problem, although it still makes the apparatus bigger, more expensive and more power hungry. However, in for example an application-specific integrated circuit (ASIC) fabrication, the number of logic gates which can be used for the whole circuit is strictly limited, so it is very undesirable to use large numbers of gates simply to fabricate local registers for the vertical interpolator.

There is therefore a conflict in this type of ASIC design between the limited number of logic gates available and the need for high quality video processing.

Accordingly the invention also provides video signal processing apparatus comprising:

a vertical interpolator operable to generate video lines of an output video signal from an input video signal, the vertical interpolator generating each output line using a respective set of interpolation coefficients dependent on the vertical position of that output line with respect to lines of the input video signal;

other video signal processing devices operable to perform video signal processing in accordance with video signal parameter data stored in local registers associated with each device;

control logic for supplying the video signal parameter data to the other signal processing devices once per video field of the output video signal and for supplying a set of interpolation coefficients to the vertical interpolator once per video line of the output video signal, each supplied set of interpolation coefficients being appropriate for the vertical position of the next-generated output video line.

The invention addresses and alleviates the conflict described above by providing for signal processing devices other than the vertical interpolator to be updated with fresh coefficient or configuration data once per field, and for the vertical interpolator to be updated with an appropriate set of interpolation coefficients once per line. This means that the vertical interpolator need only store one or two sets of coefficients (depending on when the update is made--preferably during line blanking so that only one set needs to be stored) instead of the large number (e.g. 32) of sets previously required.

The invention therefore also recognises that the storage problems associated with the vertical interpolator tend not to apply to other processing devices in this type of apparatus (e.g. a horizontal interpolator), since they do not tend to require so many sets of coefficients. The line-by-line update need not therefore be applied to the other devices.

The vertical interpolator coefficients need be updates only once per output active line, although they could be updated more often if this is more convenient--e.g. once per input line.

The above and other objects, features and advantages of the invention will be apparent from the following detailed description of illustrative embodiments which is to be read in connection with the accompanying drawings, in which:

FIG. 1 is a schematic diagram of a down-conversion apparatus;

FIG. 2 is a schematic diagram of an input and horizontal interpolation stage of the apparatus of FIG. 1;

FIG. 3 is a schematic diagram illustrating coefficient address generation by a vertical interpolator;

FIG. 4 is a schematic diagram of a control apparatus;

FIG. 5 is a schematic diagram of an image enhancer;

FIG. 6 is a schematic diagram of a gamma converter;

FIG. 7 is a schematic graph illustrating the characteristics of the converter of FIG. 6;

FIG. 8 is a schematic diagram of an inverse gamma converter;

FIG. 9 is a schematic graph illustrating the characteristics of the converter of FIG. 8;

FIG. 10 schematically illustrates the contents of a control read only memory (ROM) in the control apparatus of FIG. 4;

FIG. 11 schematically illustrates the contents of a control random access memory (RAM) in the control apparatus of FIG. 4;

FIG. 12 is a schematic timing diagram illustrating an active data enable signal;

FIGS. 13a and 13b are schematic diagrams illustrating the operation of a vertical interpolator of the apparatus of FIG. 1;

FIG. 14 schematically illustrates a video camera incorporating the apparatus of FIG. 1;

FIG. 15 schematically illustrates a video camera connected to the apparatus of FIG. 1; and

FIG. 16 schematically illustrates a video tape recorder incorporating the apparatus of FIG. 1.

FIG. 1 is a schematic diagram of a down-conversion apparatus. All of the apparatus of FIG. 1, with the exception a FIFO memory 150, is embodied as a single application-specific integrated circuit (ASIC) 10.

The down-conversion apparatus receives a high definition digital video signal at its input and generates a standard or conventional definition digital video signal at its output. The down-conversion process involves horizontal down-conversion followed by vertical down-conversion. No temporal conversion is performed as both the input and output video signals are assumed to be at the same frame or field rate. The spatial down-conversion is followed by image enhancement, color mapping and formatting of the video data to the required output standard.

Various different standards conversions can be performed by the apparatus of FIG. 1, under the control of configuration data to be described below. These conversion modes are:

1035 lines/2:1 interlaced to 485 lines/2:1 interlaced

1080 lines/2:1 interlaced to 485 lines/2:1 interlaced

1035 lines/2:1 interlaced to 364 lines/2:1 interlaced

1080 lines/2:1 interlaced to 364 lines/2:1 interlaced

1152 lines/2:1 interlaced to 576 lines/2:1 interlaced

1035 lines/2:1 interlaced to 485 lines/1:1 progressive scan

1080 lines/2:1 interlaced to 485 lines/1:1 progressive scan

High definition input video data is passed first to a half-band luminance filter 20 and from there to a horizontal interpolation stage 30. These stages are illustrated in more detail in FIG. 2, which shows the way in which a luminance data stream (Y) and a multiplexed chrominance data stream (PbPr) are received and treated differently at the input stages of the down converter. In particular, only the luminance data stream is supplied to the half-band filter 20. The multiplexed chrominance data stream is first passed to a demultiplexer 22 and the demultiplexed Pb and Pr signals are both delayed by delay elements 24 to compensate for the delay of the half-band horizontal filter. The filtered luminance data and delayed demultiplexed chrominance data are then passed to three respective horizontal interpolators 30', 30" and 30'".

The justification behind this initial stage of processing is that in a so-called 4:2:2 high definition digital video signal at, say, a standard of 1125 lines, the luminance data stream (Y) has a data rate of 74 MHz and the two chrominance data streams Pb and Pr have respective data rates of 37 MHz (when these are multiplexed together, they provide a composite chrominance data stream having a data rate of 74 MHz). Because integrated circuits operating on a data stream at 74 MHz are technologically more difficult to produce and generally have a higher power consumption than circuits for handling data at 37 MHz, the first operation performed on the luminance data is to half-band filter and (not shown) decimate the data to provide a luminance data rate of 37 MHz. This means that only the half-band horizontal filter 20 in the luminance data path has to operate at a data rate of 74 MHz. (In order to operate at such a high processing speed, the half-band horizontal filter 20 is implemented using a shift and add architecture rather than a multiply and add architecture.)

Similarly, the 74 MHz multiplexed chrominance data stream PbPr is first demultiplexed (by the demultiplexer 22) to provide two separate 37 MHz data streams for subsequent processing. This means that in the chrominance data path only the demultiplexer 22 has to operate at the data rate of 74 MHz.

This arrangement also has the advantage that after half-band filtering and decimation of the luminance data, all subsequent processing takes place on luminance and chrominance data streams having identical data rates, i.e. on a 4:4:4 video signal. This provides the flexibility to generate an output signal in either the 4:4:4 standard or a 4:2:2 standard.

A further advantage of this arrangement is that the three horizontal interpolators 30', 30" and 30'" can all be made identical. In the case of an ASIC design, this can simplify the design work tremendously.

Returning to FIG. 1, it should be noted that only one of the three parallel signal paths (for the Y, Pb and Pr data respectively) is illustrated, for clarity of the diagram.

The purpose of the horizontal interpolator 30 is to convert from the number of pixels per line of the input (high definition) standard to the required number of pixels per line in the output (standard definition) standard. The horizontal interpolator 30 provides two ways of achieving this, to be referred to as "squeeze" and "edge crop" modes.

In the squeeze mode, the image content of an entire line of input pixels is used to generate a respective line of output pixels. This may lead to some horizontal compression distortion of the image if the aspect ratio of the output video signal is different to that of the input video signal. In contrast, in the edge crop mode, some image content at one or both ends of each line is discarded when the output lines are generated, but this has the compensation that horizontal compression distortion is not introduced into the output picture if the difference in input and output aspect ratios is compensated by the correct amount of edge cropping.

Therefore, the horizontal interpolator 30 has to generate lines of output pixels based on either a subset or all of each line of input pixels. In the edge crop mode it is capable of providing one quarter pixel offsets for fine adjustment of the horizontal position of the output (cropped) image with respective to each input image. In the squeeze mode it simply reduces the sample rate of the lines of pixels, generally by a factor of about 3/4.

The vertical conversion process involves vertical interpolation followed by line dropping.

The output of the horizontal interpolator 30 is passed to a 12 line delay device 40 and from there to a 13 tap vertical interpolator 50. This generates output lines of pixels at the same line rate as the input lines. However, as described below, only some of the output lines of the vertical interpolator are treated as "active" lines, and the remainder are "dummy" lines which are discarded.

An optionally connected image enhancer 60 is provided at the output of the vertical interpolator. The image enhancer will be described in more detail below with reference to FIG. 5.

The video data output by the image enhancer is passed to a conventional YC (luminance-chrominance) to RGB (red-green-blue) matrix 70. This generates red, green and blue component video signals at its output. The component video signals (or a color bar test signal provided by a color bar test generator 80) are supplied to an inverse gamma converter 90 which applies an inverse gamma correction function, followed by a colorimetry converter 100 and a gamma converter 110. The purpose of the chain of the inverse gamma converter 90, the colorimetry converter 100 and the gamma converter 110 is firstly to allow any gamma conversion performed in accordance with the high definition format to be reversed, then for the red, green and blue color standards to be changed from those applicable to the high definition format to those for a standard definition format, and finally for gamma correction appropriate to the standard definition format to be applied. (It is noted that the gamma correction and color or phosphor standards associated with high definition and standard definition video are usually different.) This arrangement also allows the colorimetry conversion to be performed on the basic R, G and B components rather than the non-linearly processed gamma-corrected versions of those components.

The output of the gamma converter 110 is supplied to an external FIFO memory 150. The purpose of the FIFO memory 150 is to allow data to be input at the clock rate associated with the input video signal and to be output at the clock rate associated with the output video signal. Thus, in this case, data is written into the FIFO memory at a clock rate related to the data rate of the input video signal, and is read out at a clock rate related to the required output video signal format.

Also by providing the FIFO memory 150 as an external connection at this point in the circuit, an external colorimetry converter or image enhancer can be inserted into the processing stream using the connections provided for the FIFO memory 150.

The output of the FIFO memory 150 is supplied back to the ASIC 10 and is routed through an optionally connected RGB to YC matrix 120 which allows the standard definition output to be in either RGB or YC form, and an optionally connected half-band chrominance filter 130. If the filter 130 is out of circuit, the output is in a 4:4:4 format. If the filter 130 is in circuit, the two chrominance data streams are half-band filtered and sub-sampled so that the output is in a 4:2:2 format. Finally, the signal is passed to an output multiplexer and blanking inserter 140.

FIG. 3 is a schematic diagram illustrating the operation of the vertical interpolator 50.

In general, an output line will not coincide spatially with the vertical position of an input line. Depending on the relative position of the current output line with respect to the input lines, a different set of coefficients will be required for the filtering process performed by the vertical interpolator. In the present system, 32 possible sets of coefficients are provided, so that the vertical position of an output line can be specified to an accuracy of 1/32 of the input line spacing.

The 32 sets of coefficients are not stored at the vertical interpolator, but instead the appropriate set for each output line is loaded into registers 52 at the vertical interpolator during the line blanking period immediately preceding each line. This avoids the need for a bank of 32 sets of 13 ten-bit registers at the vertical interpolator to store the full set of coefficients which may be required during generation of an output field.

Therefore, in order to select the correct set of coefficients to be loaded into the vertical interpolator registers at the line blanking interval, the vertical interpolator comprises a counter 54 which maintains a cyclic count of the position of the next output line with respect to the input lines, expressed, in this case, in modulo 32 arithmetic to correspond to the 32 possible sets of coefficients. This address signal is transmitted to control logic (not shown in FIG. 1, but provided within the ASIC 10 and schematically illustrated in FIG. 4) to specify the correct set of coefficients to be transmitted to the vertical interpolator at the next line blanking interval.

FIG. 3 illustrates a simplified version of this process, in which three possible address values (A=0, 1, 2) may be used, corresponding to three possible sets of filter coefficients. It will be seen from FIG. 3 that the address value transmitted to the control logic depends on the relative vertical position of the current output line with respect to the input lines.

The control logic which handles the updating of the vertical interpolator coefficients and many other control matters is illustrated schematically in FIG. 4.

The control logic comprises two memories for storing coefficients and other configuration data associated with the various devices illustrated in FIG. 1. These are a dual-port random access memory (RAM) 200 and a read only memory (ROM) 210. The writing and reading of data in the dual-port RAM 200, and the reading of data in the ROM 210, is controlled by an address generator 220.

The contents of the dual-port RAM 200 and the ROM 210 will be described in more detail below, but briefly the ROM 210 contains a respective set of horizontal and vertical filter coefficients for each possible mode of operation of the down converter (as listed above), together with various possible configuration data for devices such as the image enhancer and the colorimetry converter. The ROM 210 also contains a 10-byte configuration word which defines a default mode of operation of the down converter.

The dual-port RAM 200 mirrors the structure of the ROM 210 in that it provides storage capacity for one set of horizontal and vertical filter coefficients, one set of configuration data for the image enhancer, colorimetry converter, etc., and a 10-byte configuration word.

In operation, if the default mode of operation (e.g. 1125 line, 1035 active lines high definition input to 525 line output conversion, interlaced, squeeze mode) is required, it is not necessary to load anything into the dual-port RAM 200. The use of these various sets of coefficients is specified by the 10-byte configuration word permanently written into the ROM 210, and the required coefficients are found in the ROM 210.

If, however, a different mode of operation is required but using alternative coefficients still held in the ROM 210, it is merely necessary to write a different version of the 10-byte configuration word into the RAM 200. This will then provide pointers to non-default sets of coefficients and configuration data stored in the ROM 210.

In a further possibility, if a mode of operation is required which cannot use the possible sets of coefficients and other data stored in the ROM 210, replacement coefficients can be stored in the RAM 200 and pointed to by the 10-byte configuration word held in the RAM 200.

Data can be loaded into the RAM by either a serial or a parallel interface (230, 240 respectively). Data read from the RAM and the ROM are supplied to a multiplexer 250 which selects either RAM-stored data or ROM-stored data in dependence on the state of the 10-byte configuration word either as held in the ROM or as overruled by that held in the RAM.

Each of the modules, including the horizontal and vertical interpolators, of FIG. 1 is connected by a common data and address bus. When it is required to update configuration or coefficient data relating to one of the modules, an appropriate address is placed onto the address bus by the control logic and the corresponding data placed onto the data bus. In general, coefficient and control data are updated during field blanking, with the exception of vertical interpolator coefficients which are updated during line blanking.

FIG. 5 is a schematic diagram of the image enhancer 60. The image enhancer's basic function is to amplify high frequency components of the video signal to enhance the reproduction of image detail. Techniques very similar to those used in, for example, the Sony DVW-700/700P and BVW-D600/D600P camera-recorders are employed.

The processing is applied only to the luminance signal; the chrominance components are simply subjected to compensating delays 300.

The luminance signal Y is therefore supplied in parallel to a delay unit 310 and a level dependent processor 320. This applies a non-linear level dependence function to the luminance signal which is designed to reduce the enhancement in dark areas which, due to gamma correction tends to contain more noise than lighter areas. The function is achieved by clipping the input pixel values to, say, 192 (out of a possible range of values up to 1024 or 10-bits), scaling the clipped signal by a factor such as 1/16, truncating, and subtracting this scaled signal from the original input signal.

There follows high-pass and band-pass filtering in the horizontal and vertically directions to provide an enhancement of detail (high frequency) components and also to avoid so-called cross-color when the down converted video is ultimately encoded using the NTSC or PAL standards.

The ratio of horizontal and vertical enhancement can be adjusted by applying a horizontal/vertical detail ratio to a multiplier 330. The combined signal is then subject to another non-linear function known as "crispening". This is designed to reduce the amount of detail added to small objects and also to reduce the amount of noise in the detailed signal. This is achieved by setting a threshold on the detail signal below which the detail gain is zero. Another function of this block is to clip the detail signal so as to limit the maximum amount of detail that may be added. The function can be achieved by clipping the input to a variable level and then subtracting the clipped signal from the original.

The level of the detail signal can be adjusted by a further multiplier 340 before it is added (350) to the delayed original luminance signal.

FIGS. 6 and 8 are schematic diagrams of the gamma converter and inverse gamma converter respectively, and FIGS. 7 and 9 illustrate the required gamma conversion and inverse gamma conversion transfer functions.

In each of the gamma converter and the inverse gamma converter a 64-value look-up table (LUT) 400 (in FIG. 6) is addressed by the six most significant bits of each pixel value. This yields a difference signal and a base signal as outputs. The seven least significant bits of each input pixel value are multiplied by the difference signal and are then added to the base signal to generate an output pixel value. This processing is applied in parallel to the red, green and blue pixel values. The difference and base values are selected to approximate to the required curves shown in FIGS. 7 and 9.

FIG. 10 schematically illustrates the contents of the control ROM 210.

The ROM 210 stores the ten-byte configuration word in its first ten bytes 0.times.000 to 0.times.009, followed by:

two sets of horizontal interpolation coefficients (for edge crop and squeeze mode respectively);

four sets of vertical interpolation coefficients;

two sets of inverse gamma coefficients;

two sets of output gamma correction coefficients; and

three sets of colorimetry conversion coefficients.

The ten-byte configuration word stored in the ROM 210 specifies one default mode of operation of the 2.times.4.times.2.times.2.times.3 permutations of the sets of coefficients listed above.

If a non-default mode of operation is required, but still using coefficients stored in the ROM 210, then all that is necessary is to load a replacement ten-byte configuration word into corresponding addresses in the RAM 200, as shown in FIG. 11.

If further variations from the stored modes of operation are required, then further coefficients can be loaded into respective address ranges in the RAM 200 (see FIG. 11), and their use specified by setting the RAM's ten-byte configuration word accordingly.

The choice of whether the RAM's or the ROM's configuration word is to be used is set by detecting whether either a serial or a parallel external data transfer port to the dual port RAM is enabled (in which case the RAM's configuration word is to be used), or by an external control input to the control logic (not shown). However, in another embodiment the choice could be set by detecting whether a configuration word has been written into the RAM since the apparatus was last powered up.

FIG. 12 is a schematic timing diagram illustrating the use of an active data enable signal in the apparatus of FIG. 1.

As mentioned above, the vertical interpolator produces one line of pixels at its output for every line of pixels at its input. Similarly, the horizontal interpolator generates output pixel values at the same data rate as the pixel values at its input.

The reason for arranging the circuit in this way is that all of the processing devices from the horizontal interpolator to just before the external FIFO memory can operate at the same clock speed; there is no need for a different clock at the output of the horizontal interpolator, and then a different clock again at the output of the vertical interpolator.

However, the arrangement does mean that some of the pixel values output by the horizontal interpolator are "dummy" values which are not required in the output image, and similarly some of the lines of pixels output by the vertical interpolator are completely "dummy" lines.

Therefore, in order to distinguish between real and dummy data values, an "active enable" signal is employed. This is generated by counters (not shown) associated with the horizontal and vertical interpolators, so that if, for example, the horizontal interpolator is working at a sub-sampling ratio of 3/4,then the active enable flag will be high (indicating dummy data) for one in every four pixel values output by the horizontal interpolator. Similarly, if the vertical conversion ratio is about 1/2,then approximately one half of the lines generated by the vertical interpolator will be flagged as "dummy" lines. Overall, this will give only about 3/8 of the pixel values output by the horizontal and vertical interpolators being flagged as "active".

The active enable signal is actually generated by a logical OR combination of the respective active enable signals generated in respect of horizontal and vertical interpolation.

At each processing stage in the apparatus of FIG. 1 between the horizontal interpolator and the external FIFO, input data is latched (sampled) as a first operation, in accordance with a common clock signal. The active enable signal is supplied in parallel (i.e. not latched or delayed) to each device. If the active enable signal is high, then quite simply the current input to each device is not latched and the existing value maintained. This has the effect of "stretching" certain clock cycles for the processing elements following the horizontal interpolator.

This process is illustrated in FIG. 12, in which DO, D1, D2 . . . D7 are active pixel values, and X signifies a dummy value at every fourth pixel output by the horizontal interpolator. The second line of FIG. 12 shows the active enable signal.

When the data output by the horizontal interpolator is latched by the next device to receive it, the active enable signal prevents the dummy value X from being latched, and instead the values D2 and D5 are held for two clock cycles and processing pauses. Similarly, at the next line, the values D1 and D4 are held for two clock cycles.

FIGS. 13a and 13b schematically illustrate part of the operation of the vertical interpolator.

In one possible use of the apparatus of FIG. 1, the input high definition video signal could be derived from a high definition source operating at a field rate of 60 Hz. However, if an NTSC output is required, the field rate must be altered by one part in one thousand to 59.94 Hz. This can conveniently be done using a fielddropping converter upstream of the down-conversion apparatus, which drops one field in every few seconds of video, but this leads to sudden inversions in the field polarity of the signal supplied to the down-converter.

The effect of a polarity inversion is particularly evident in a video format having an odd number of active lines, since one field will have one more line than the other field.

In order to cope with occasional field polarity inversions in its input video signal, the vertical interpolator is arranged so that it can generate a particular polarity output field from either polarity of input field. This is achieved quite simply in the present design, since the vertical interpolator (in particular the counter described with reference to FIG. 3) will automatically request the correct set of interpolation coefficients for each output line in dependence on the relative position of that output line with respect to the input lines. Therefore, the only difference between an odd and an even input field as far as the requesting of interpolation coefficients is concerned is that the spatial positions of the input lines change by one half of one line.

However, it has been appreciated that this could have led to subjectively disturbing artifacts at the vertical extremes of the picture, so steps have been taken to avoid this problem.

The potential problem which has been identified and addressed is that if the larger output field (i.e. the one having the extra line) is derived from and spatially aligned with the larger input field (again, the one having the extra line), then if the input field polarity inverts, the larger output field actually extends beyond the (smaller) input field from which it must be derived. The extreme top and bottom lines of the output field are then derived at least partly from black lines in the nonactive area of the input field. However, in the other input polarity the extreme output lines are derived from active lines of the input field. Therefore, when an input field polarity inversion occurs, the extreme lines of the larger output field change in luminance because they change between an interpolation from active lines and an interpolation from non-active lines.

This problem is addressed in the present system by spatially aligning the extreme lines of the output frame (i.e. the extreme lines of the larger output field) with non-extreme lines of the larger input field (i.e. at least one line in from the top and bottom of the input frame). This means that the output picture is very slightly vertically cropped, but it is considered that this is preferable to subjectively disturbing luminance variations at the top and bottom edges of the picture every few seconds when an input polarity inversion occurs.

FIGS. 13a and 13b show a particular example of this technique. In FIGS. 13a, input fields F0 and F1 of a high definition video signal are converted to respective output fields F0 and F1 of a lower definition signal. In particular, output field F0 is derived wholly from input field F0, and output field F1 is derived wholly from input field F1. Similarly in FIG. 13b, the input fields F0' and Fl' are in the opposite field polarity. Output field FO is derived wholly from input field FO' and output field F1 is derived wholly from input field F1' .

Because the extreme top and bottom lines of the larger output field are spatially aligned with the next-to-top and next-to-bottom lines of the larger input field, the larger output field never extends beyond the input field from which it is to be derived, even if there is a field polarity change in the input video signal.

FIGS. 14, 15 and 16 schematically illustrate three applications of the apparatus of FIG. 1. In particular, FIG. 14 schematically illustrates a video camera incorporating the apparatus of FIG. 1; FIG. 15 schematically illustrates a video camera connected to the apparatus of FIG. 1; and FIG. 16 schematically illustrates a video tape recorder incorporating the apparatus of FIG. 1.

In FIG. 14, a video camera 400 comprises a high definition light sensing element 410 operable to generate a high definition video signal. A down-converter 420 of the type described above is placed in a parallel signal path (with a video analogue-to-digital converter as appropriate) so that the camera can output substantially simultaneously a standard definition and a high definition signal. If desired, a delay element could be inserted in the high definition signal path to compensate for the processing delay of the down converter 420.

FIG. 15 illustrates a similar arrangement with the camera generating a high definition 60 Hz video signal. A field dropping temporal standards converter 430 is connected between the high definition camera output and the down converter 420. the field dropping converter operates to change the field rate of the video signal from 60Hz to 59.94 Hz (a change of one part in 1000) by dropping a video field once in every few seconds. This leads to discontinuities in the field polarity sequence of the signal supplied to the down converter 420, but this is compensated by the operation of the down converter as described above.

Finally, FIG. 16 schematically illustrates a high definition video recorder/player 440, again incorporating a down converter 420 to provide substantially simultaneous outputs in high definition and standard definition formats.

Although illustrative embodiments of the invention have been described in detail herein with reference to the accompanying drawings, it is to be understood that the invention is not limited to those precise embodiments, and that various changes and modifications can be effected therein by one skilled in the art without departing from the scope and spirit of the invention as defined by the appended claims.