GB2114404A - Generating a colour video signal representative of a stored picture - Google Patents

Abstract

Colour video signals are generated representative of a stored picture, e.g. for use with an electronic tablet or graphics system, by reading from a code store (20) a succession of colour codes each indicating a pixel colour, looking up in a palette store (22) the colour component values corresponding to each colour code, and combining in an interpolator (30) two selected colour component values to provide an output. The combination function is determined by a circuit (32) on a pixel-by- pixel basis from control codes included in the sequence of codes stored in the code store (20). The interpolated colours may be not available on the palette, and if the interpolation is to improve the appearance of a colour transition, the position of the edge may be defined to a fraction of a pixel. <IMAGE>

Description

SPECIFICATION Generating a colour video signal representative of a stored picture This invention relates to a method and apparatus for providing a video signal representative of a stored picture. An example of such apparatus is an electronic graphics systems, as used by a graphic artist.

A conventional electronic graphics system for composing pictures for display on a television screen comprises a frame store in which each storage position corresponds to an element of the picture, termed a pixel, and the number stored in that position corresponds to the colour and brightness of that pixel. For convenience we shall use the term 'colour' to include also 'colour/brightness combination'.

Typically, an eight-bit binary number will be stored for each pixel, allowing 256 possible colours to be called from a store known as the 'palette'. Although this 'painting-by-numbers' approach gives useful results, a boundary between two coloured areas will often exhibit unwanted effects due to the necessity for each pixel to be either one colour or the other, resulting in a quantisation effect sometimes known as 'jagging'. This is particularly apparent when a straight boundary is nearly horizontal or nearly vertical, giving rise to a stepped edge.

We have appreciated that in order to depict smooth edges it is necessary to provide intermediate colours and brightness. The 'painting-by-numbers' system can do this, but with only 256 numbers available the range of colours becomes severely restricted because so much of the storage is taken up by the intermediate colours needed at the colour boundaries. Alternatively the storage requirement must be enormously increased.

It is well known for certain purposes to combine adjacent or related television lines in specified proportions. Examples are with television standards conversion, or in the system of UK Patent Application 2040649A. However, combining lines in this way does not overcome the problem of 'jagging' (except in very specific circumstances), and causes a loss of resolution over the whole width of the picture.

The problem with the electronic graphics system described is one particular example of a more general problem, namely to improve the quality of the display of a stored video signal without a directly proportionate increase in the cost or the required storage capacity.

The invention in its various aspects is defined in the claims below.

A preferred embodiment of the invention will now be described by way of example with reference to the drawing, in which: Figure lisa block diagram of an electronic graphics system embodying the invention, and Figure 2 illustrates one possible combination law for combining two colours for a series of pixels.

The system illustrated is designed in particular for use with carton-like still television pictures made up of areas of uniform colour which meet at boundaries (edges). In particular this includes text and lettering on a plain background. The edges can have any angle relative to the line scan direction (as exemplified by a circle), and the sharpness of the edge can ideally range from the sharp limit set by the capabilities of the video system, namely the horizontal bandwidth and vertical line structure, to an 'edge' which is wider than the picture itself and appears as a smooth colour graduation or colour wash.

In the illustrated example the problem of jagging at edges in the picture is overcome by generating in the boundary region, colours for the individual pixels (picture elements) which are intermediate between the colours of the areas adjoining the edge. When a sequence of intermediate colours is filtered at the output of the apparatus it can produce an edge whose position can be varied by fractions of a pixel spacing and whose width is several pixels. It will be appreciated that there are relatively few edges in the picture compared to the number of pixels. The result obtained is comparable to that obtained when a television camera is used to capture the image.

Figure 1 shows an electronic graphics system 10 embodying the invention. The system includes one or more input devices 12 which may each take the form for example of a sensitive graphics tablet. This has a surface across which a stylus can be drawn and generates output signals representing in two co-ordinate form the instantaneous position at which the stylus is pressed on the tablet. The output of the input device is applied to control circuitry 14, which is connected to receive scanning signals from timing, scanning and display circuitry 16. A picture store 20 is connected to the control logic 14 and the timing circuitry 16 and has a capacity to store a colour code indicator for each pixel.

The picture store 20 has its output connected to a "palette" 22, which is a store which converts the codes stored in the picture store 20 into components representing the red, green and blue signal components required to produce the desired colour.

Essentially the palette performs the function of a look-up table. The numbers in the picture store 20 and in the palette 22 can be altered by the control logic 14, under the control of the input device 12. The palette output is applied to output digital-to-analogue converters 24 which provide red, green and blue output signals.

As thus far described the system is typical of a conventional "painting-by-numbers" electronic graphics generator.

In accordance with this invention, the system includes additional components as follows. The palette 22 is not directly connected to the output converters 24, but instead is connected to two registers 26 and 28 which can each hold the red, green and blue component values for one of the palette colours. The registers are both connected to an interpolator 30 which can combine the contents of the two registers in desired proportions.

The proportions are determined by a code interpreter circuit 32 which receives information of the desired combination function and holds this as a combination function for each pixel. The code interpreter controls the interpollator 30 for combination in accordance with this function. The interpreter 32 in fact provides two outputs (1 - S) and S which are associated with the outputs of the registers 26 and 28 respectively.

Output filters 34 are desirably included for the red, green and blue analogue outputs respectively.

Thus, the output red, green and blue signal for any pixel is a function of two stored colours A and B from the palette and the variable S. Typically the three components R,, G,, BO of the output will be a linear mixture of the corresponding components of the stored colours A and B: (1 - S) . RA + S.RB; (1 - S).GA + S.GB; (1 - S) . BA + S.Bs where S varies between 0 and 1.

One method of operation of the system illustrated will now be described. The picture store 20 accommodates one eight-bit code for each pixel. The palette 22 stores the colour components of up to 128 colours, thus requiring seven bits. The additional bit associated with each pixel is a control bit. The interpolation coefficient S has up to seven bits.

At the start of each television line reset circuitry (not shown) sets the value of the interpolation coefficient S stored in the code interpolator 32 to zero, and loads the palette colour corresponding to colour code 0000000 as colour A into register 26 and the palette colour corresponding to colour code 1111111 as colour B into register 28. The coefficient S is held at its current value until updated.

The codes for the pixels of a selected video line are then sequentially addressed in the picture store 20. The first bit of each 8-bit stored number is the control bit. If the control bit is 1, then the stored number represents a colour code and is applied to the palette 22 to select the corresponding colour component values. These values are applied to one of the registers 26,28, the selected register being identified by the value of the most significant bit of the interpolator coefficient S, ie whether S is more or less than one-half. If this bit has value 1, then the register 26 is used, if 0 then register 28, i.e. it is applied to the register currently constituting the minor portion of the output.If the control bit of the 8-bit number read from the picture store 20 is zero, then the stored code is not a colour code but instead represents a new value for the interpolation coefficient S. The code interpreter 32 in Figure 1 converts the stream of numbers from the picture store 20 into commands such as "load colour register A (or B) with colour N from the palette" and "change the interpolation coefficient S to M". The coded instructions for the code interpreter are entered into the picture store by the control logic 14. The detailed operation is most easily explained by means of a simple trivial example.

It will be assumed that the video line considered starts with a green portion, has a sharp boundary into a blue portion, and a slightly more diffuse boundary into a red portion. The colour codes for green, blue and red are assumed to be 0001000, 0001100 and 0001110 respectively. The sequence of stored codes read from the picture store 20 will be as follows.

Pixel Code from Value of inter- Colour codes of colour Remarks No store 20 polatorco- componentvaluesheld efficientsheld in: in code inter polator32 Register 26 Register 28 - - 000 0000 000 0000 1111111 Initial state 1 10001000 000 0000 000 1000 1111111 Load green into register 26 2 0 000 0000 000 0000 000 1000 1111111 Display green 101 10001100 1111111 0001000 0001100 loadblueinto register 28 102 00000000 0000000 0001000 0001100 continuetodis play green 121 00100000 0100000 0001000 0001100 Firstboundary- Display 25% blue 75% green 122 01000000 1000000 0001000 0001100 Display 50% blue 50% green (cyan) 123 0 110 0000 110 0000 000 1000 000 1100 Display 75% blue 25% green 124 01111111 1111111 1000 000 1100 Display 100% blue 201 10001110 1111111 0001110 0001100 load red into register 26 202 01111111 1111111 0001110 0001100 Continuetodis play blue 221 01110000 1110000 0001110 0001100 Second boundary Display steadily 222 01100000 1100000 0001110 000 1100 increasing proportions of 223 01010000 1010000 0001110 0001100 red to blue 224 0 100 0000 100 0000 000 1110 000 1100 225 00110000 0110000 0001110 0001100 226 00100000 0100000 0001110 0001100 227 00010000 001 0000 0001110 0001100 228 0 000 0000 000 0000 000 1110 000 1100 Display 100% red Thus it is seen that some of the codes read from the store (those beginning 1) serve to identify one of the available coiours for which colour component values are held in the palette store 22, while the other codes (those beginning 0) are indicative of the manner in which the colours for which colour component values are held are to be combined to produce intermediate colours. When these control codes are outputted, the palette colours from the previous pixel are used for the current display, and it is not possible to access a further one of the colours in the palette. However, we have appreciated that with the type of artificial picture we are here concerned with, which has large areas of uniform colour, this is not in fact a problem.

It should be noted that colours intermediate between the palette colours may persist for many output samples to represent an intermediate colour in a nearly horizontal transition.

Furthermore the system can be used to provide colours intermediate the palette colours for other purposes, not related to transitions between two areas of uniform colour. In particular, the system extends generally the range of colours to include colours which are between any pair of palette colours. Forthis purpose the desired combination function will remain constant over the areas of intermediate colour. Thus, instead of providing 256 colours alone, it is now possible to provide 128 colours directly from the palette, and for each pair of these 128 colours to provide 128 shades of colour formed by linear combination of the colours of the pair.

In an alternative mode of operation the coefficient S is not stored directly in the picture store 20, but is selected indirectly by means of parameters which are set in the picture store. In this instance the basic shape of a transition curve as shown in Figure 2 is stored in the control logic 14 and is available to the code interpreter. As stored, the curve is not associated with a particular timescale, that is to say slope, or sharpness of transition. To identify a transition, therefore, two factors have to be specified (1) its sharpness and (2) its position along the line conveniently identified by the position of the mid-point of the transition. To gain the full benefit the edge should be positioned to an accuracy of a fraction of the pixel spacing. In the example of Figure 2 the edge has a rise-time of six pixel spacings and is thus sampled at six points.The "phase" of the samples is related to the precise position of the edge. As shown, the centre of the edge is one quarter of the way from one pixel to the next.

In this alternative method, 255 colour codes are available and the corresponding component values are stored in the palette store. The remaining code is reserved as a unique "escape" code. The code interpreter 32 monitors the output of the picture store 20, and when it detects an escape code treats the next code as an 8-bit control code and inhibits the registers 26 and 28 from accepting the colour component values from the palette for that pixel, thus holding the values from the pixel prior to the escape code.The control code describes the nature of the transition to the new colour as one of 256 combinations of slope (i.e. abruptness of transition) and position (relative to the picture element grid). For example, one of 16 slopes, varying from one to ten pixels between the 10% and 90% points, could be combined with a centre of the transition specified to an accuracy of one-sixteenth of the pixel spacing. The receipt of the control code initiates the transition. Thus the escape and control codes must be stored at an appropriate distance prior to the mid-point of the edge, where the transition is to begin.

The commands for the colour A and B stores 26,28 in Figure 1 and the interpolation coefficient S, are stored in the code interpreter 32. The repertoire of commands can be fixed, or programmable in the same way that the contents of the palette 22 can be programmable.

As a further modification, the escape code can be replaced by an escape sequence of two codes, corresponding to two pixels. One, say the first, would indicate the nature of the edge to be synthesised, and the other would give the code number of the new colour. If the "next" colour were set up as described in the first example above, the "next" colour can then become the current colour after the edge has been passed.

Then unless a new "next" colourwere defined, the "old" colourwould automatically become the new "next" colour. This simplifies the operation when, for example, displaying text in one colour against another background colour, where the two colours alternate a number of times along the video line.

The type of picture which has been discussed comprises areas of uniform colour separated by boundaries.

In composing the picture the boundaries can be stored, to an accuracy better than the nearest pixel, in terms of the position and slope of each edge as it intersects the television scanning lines. This information comes from the co-ordinate pairs produced by the graphics tablet, together with knowledge of the shape of the boundary (e.g. circle, polygon, cursive freehand) selected by the artist. The colour codes can then be written at appropriate points to set up the colours for the "next" areas along the line.

Even when all this has been done, most of the store, corresponding to large areas of uniform colour, will not contain any information. In practice it could be filled with a repeat of the colour control code, a "null" code, or an escape code enabling further control operations to be provided.

The circuitry shown in Figure 1 has been shown as discrete circuitry. However, it will be appreciated that in practice at least the greater part of the functions can be implemented on a general purpose computer, and in this case Figure 1 should be regarded as a flow chart or logic diagram.

The technique described is expected to be of value in improving the end product of electronic graphics generators. It is also potentially useful as a method of improving the quality of the output from an enhanced teletext system, where coding based on picture description instructions is used. In such a system the coded pictures could be transmitted as a sequence of codes, in which case the large redundant areas corresponding to uniform colour can be removed by a form of run-length coding or direct addressing.

Alternatively, however, instructions could be transmitted, e.g. of the form "draw a circle radius r centre x,y".

Other uses of the apparatus may also be apparent. In particular for such other uses the apparatus may be modified, for example, the picture store 20 may in fact only store a part of a complete television picture over which the system is to be used. This might apply where the system is used for captions or subtitles at the bottom of the picture for example.

Claims (5)

1. Apparatus for providing colour video signals representative of a stored picture, comprising: signal storage means for storing colour codes each indicating at least one pixel colour of a picture or picture portion; palette means for storing an indication of the colour component values corresponding to each of a plurality of available colour codes; combination function storage means for storing a desired combination function; and combining means connected to the function storage means to be controlled thereby selectively on a pixel-by-pixel basis to combine two colour component values from the palette means in desired proportions for a given pixel in accordance with the stored combination function to provide an output signal.

2. Apparatus according to claim 1, in which the function storage means is connected to the signal storage means to receive combination function information.

3. A method of electronically generating colour video signals representative of a stored picture, comprising: defining a desired combination function; reading from store a succession of colour codes each indicating at least one pixel colour of a picture or picture portion; reading from store an indication of the colour component values corresponding to each of the colour codes; and combining two colour component values to provide an output signal for one pixel, the combination being controlled by the combination function selectively on a pixel-by-pixel basis.

4. A method according to claim 3, in which the succession of colour codes includes control codes indicative of the nature of the combination function.

5. An encoded colour video signal, which includes a succession of codes associated with successive pixels with some codes being colour codes identifying one of a plurality of available principal colours and other codes being control codes indicative of the manner in which principal colours are to be combined to provide intermediate colours.