GB2293511A - Colour television decoder using a 3D filter - Google Patents

Abstract

The present invention relates to a colour television decoder comprising a continuously variable two or three dimensional finite impulse response filter using both spatial dimensions (horizontal and vertical) and the temporal dimension simultaneously. <IMAGE>

Description

Colour Television Decoder European Patent GB 86222680 describes an invention relating to a colour television decoder.

According to that invention there is provided a colour television decoder comprising a continuously variable two dimensional finite impulse response filter having: an input for receiving an input colour video signal representing picture information; an output for supplying chrominance signals; first delay means coupled to said input for providing a plurality of signals and first variable combining means for combining the plurality of signals and said decoder further comprising first control means for continuously controlling the first variable combining means by determining the proportions of the plurality of signals to be combined in accordance with the picture information, characterised in that the plurality of signals provided by said first delay means represent a two dimensional pattern of picture points and said first variable combining means combines said plurality of signals linearly.

GB 86222680 continues to describe preferred types of the variable filter and first control means mentioned above.

This invention describes three extensions to the invention described in 86222680.

FIRST EXTENSION: GB 86222680 describes a first delay means that provides a plurality of signals that represent a two dimensional pattern of picture points. It would be obvious to an expert in the field, that the dimensions may both be spatial, or one may be temporal.

The temporal dimension being achieved by the use of frame period delays rather than line period. This invention extends the technique to a three dimensional finite impulse response filter using both spatial dimensions (horizontal and vertical) and the temporal dimension simultaneously.

According to this invention, there is provided a colour television decoder comprising a continuously variable two or three dimensional finite impulse response filter having: an input for receiving an input colour video signal representing picture information; an output for supplying chrominance signals; first delay means coupled to said input for providing a plurality of signals and first variable combining means for combining the plurality of signals and said decoder further comprising first control means for continuously controlling the first variable combining means by determining the proportions of the plurality of signals to be combined in accordance with the picture information, characterised in that the plurality of signals provided by said first delay means represent a two dimensional pattern of picture points from one or a plurality of frames and said first variable combining means combines said plurality of signals linearly. Otherwise this extension is as described in GB86222680, including neither one or both the following two extensions.

SECOND EXTENSION: The colour information in the PAL or NTSC system is transmitted as modulation of a subcarrier signal added to the luminance signal. The modulated colour information is herein referred to as the chrominance. This subcarrier is cyclic and for constant colour information, the chrominance repeats after each integral subcarrier cycle.

In many television colour systems the phase of the colour subcarrier repeats identically after an integral number of line or frame periods (provided that the colour information being represented is the same). An example of such a system is the American NTSC system which repeats every two line periods and every two frame periods. Another example is the United Kingdoms PAL system which repeats every four line periods and every four frame periods.

This invention describes a method of using this phenomena to produce a control means with desirable properties that could be used on its own or combined with another control means, for example those described in GB86222680, to produce a superior control means. Such a control means would have both the desirable property that it detects diagonal frequencies as described in claim 11 of GB 86222680 and also the desirable property of discriminating between diagonal frequencies and colour information.

According to this invention, the first delay means described in t he above first extension and GB 86222680 would include a plurality of delays spaced so as to have identical colour video signals (provided that the colour information being represented is the same) at their outputs. A first control means would include the continuous computation of the difference or differences between the plurality of signals. It may also combine this control means with one or more others, such as those described in GB 86222680 to produce an improved control means.

This invention (the second extension to GB86222860) will be further described by means of an example. Figure 3 shows an embodiment of the invention: A PAL first control means. The input is fed to a delay line which in this example is 5 line periods long. The signals from the two taps thus created are fed to a subtractor. The output from the subtractor is rectified and smoothed by a filter. This produces a first control means that will respond to diagonal frequencies, or vertical colour transitions between the two taps, but not to constant coloured areas.

THIRD EXTENSION: The colour information in the PAL or NTSC system is transmitted as modulation of a subcarrier signal added to the luminance signal. The colour information that is used to modulate the subcarrier is represented by two components known as U and V in PAL and I and Q in NTSC. These two components are used to quadrature modulate the subcarrier. In the PAL system the V signal is inverted on every other line.

In digital television the video signal is represented by a stream of samples. Each sample represents the instantaneous magnitude of the video signal at one of a set of regularly spaced intervals. The most common interval used is one quarter of a cycle of the colour subcarrier signal. Other sampling intervals are used and it is common practice to convert these intervals to and from a quarter subcarrier by the processes of decimation and interpolation (a process often referred to as sample rate conversion) as appropriate.

When a sampling period of one quarter of a cycle of subcarrier is used there are two popular sampling phases. For example the UK PAL television system is usually represented digitally by sampling the video signal at 45 degrees to the colour subcarrier. This means that the samples are taken one eighth of a cycle after the peak of subcarrier, and at one quarter cycle intervals subsequently. This can be represented on a phasor diagram as shown in Figure 1.

The other common sampling phase is 33 degrees as used for the NTSC system.

This sampling phase can be transformed to 45 degrees by using the well known technique of interpolation.

A non PAL television signal that has been sampled with a period of one quarter cycle of subcarrier has the property that regardless of the colour being represented, there are only four different magnitudes that are ever sampled. A PAL television signal inverts the V signal on alternate lines. If the sampling phase is 45 degrees, there are still only four different magnitudes that are ever sampled.

Consider a vector corresponding to U = ul and V = vl in a PAL system Then the first sampled magnitude will be ul cos (45) + vl cos(45) Denote ul cos(45) and vl cos(45) by u and v respectively and the first sample becomes u + v, followed by u - v, -u -v, -u + v, u + v again and so on.

Because of the number of cycles of subcarrier per line there is a 90 degree shift in the subcarrier phase per line. This combined with the PAL switch means that the samples on the next line will be u + v, followed by -u + v, -u -v, u-v, u+v again and so on. These magnitudes are tabulated for a short sequence of samples on each of three lines in figure 2.

In these two special cases samples placed vertically above each other on adjacent lines often (every other sample as shown in the example) have the same sampled voltage provided that both lines represent the same luminance and chrominance information.

This invention uses this property to produce a first control means that has the desirable properties of responding to diagonal frequencies and discriminating between diagonal luminance and vertical chrominance.

According to this invention there is provided a decoder largely as described in GB 8622680 having some or all of the component parts described in GB 8622680 and also: an input for receiving an input colour video signal representing video information; a means for converting the signal to a digitally sampled signal with a sample period of one quarter of a cycle of subcarrier if the input is not already in that form; a means for converting the sampling phase to 45 degrees to subcarrier, if the input is not already sampled at that phase angle; a first control means for continuously (in this case changing frequently for example once every quarter subcarrier period) controlling the first variable combining means described in GB 8622680.

Preferably the first control means that comprises a means for synchronising to the subcarrier (this element is normally found in a decoder and is required to facilitate the demodulation of the decoded chrominance), a means for selecting those samples on adjacent lines that should have identical sampled magnitudes (if the lines represent the same luminance and chrominance) and a means of continuously computing the differences between those samples.

Preferably the first control means also includes a method of rectifying those differences and a low pass filter.

The invention will be further described by way of an example with references to accompanying drawings, in which: Figure 1 shows a phasor diagram representing the sampling of a PAL television signal. It shows the continuous variation of the subcarrier magnitude with time, arrows representing the sampling magnitudes in a digital system. The phasor diagram illustrates the phase shift on the PAL switched lines.

Figure 2 shows the sample magnitudes for a three PAL television lines of a constant colour. The shaded entries in the table illustrate those magnitudes that are the same on adjacent lines.

Figure 3 shows an example of the second extension to GB86222860, which has already been described on page 5 of this document.

Figure 4 shows a block diagram of a first control means, an embodiment of the invention. Note that the output from Figure 3 is one of the inputs of Figure 4.

The input signal is fed to a tapped delay line. The tap spacing is a line period in this example, although they could be frame period, or both. The selectors, controlled by the subcarrier phase and PAL switch synchronisation, select pairs of taps separated by one line or one frame that will have identical chrominance signals (if they represent the same colours). The subtractor, rectifier and filter produce a signal that continuously represents the difference between the samples selected. Three such difference computers are shown in this example. They are fed from taps separated by a line period, each comprising a pair of selectors, a subtractor, rectifier and filter, their outputs being combined.In this example the difference signal combiner receives inputs from both a first control means as shown in Figure 3 of this document, and a first control means of the type described in GB86222680.

Figure 5 shows a block diagram of the difference combiner referred to in the description of Figure 4.

The outputs from the difference computers all have the desirable properties of responding to most diagonal frequencies or vertical colour transitions but does not respond to true, vertically correlated colour. By combining a number of these operating over a plurality of television lines or frames reduces the likelihood that the first control means produced will fail to respond to a diagonal frequency or colour transition. In this example the performance is further improved by combining this signal with a signal of the type produced by the example in Figure 3. In this example they are combined simply by selecting the biggest of the inputs.

Finally the resulting signal is combined with a first control means of the type described in GB86222680. In this example the smallest of the two signals is selected.

This imparts the desirable property of not responding to constant colour to the first control means described in GB86222680. Because of the use of a number of difference computers and combinations with the signal from Figure 3, it does this without significantly degrading its ability to detect diagonal luminance or vertical colour transitions.

The signal thus produced is then used to control the variable combiner as described in GB86222680.

Claims (4)

1. A colour television decoder comprising a continuously variable three dimensional filter (two dimensions being spatial as in GB86222860 and the third dimension being temporal).

2. A decoder as claimed in claim 1, characterised in that the first delay means includes a first delaying circuit for providing a delay equal to one frame period and a second delaying circuit for providing a delay equal to one frame period of the input colour video signal and connected to the output of the first delaying circuit.

3. A decoder as claimed in any of the preceding claims or claims 1 to 10 of GB86222680, with first control means that continuously computes the difference between the signals at a plurality of delay taps aligned vertically or spatially, chosen so that the chrominance signals from those taps are always the same (assuming that they represent the same luminance and colour information).

4. A decoder as claimed in any of the preceding claims or claims 1 to 10 of GB 86222680, with first control means that continuously selects pairs of delay taps aligned vertically or spatially, chosen so that the sampled magnitudes of the chrominance signals from those taps are the same when selected (assuming that they represent the same luminance and colour information).