The comb filter
Recall that in TV broadcasting, the colour content of a picture is used to modulate a 4.34-MHz (3.795 MHz for NTSC) sub-carrier. The resulting side frequencies appear in clusters which fit neatly within the clusters produced by the luminance side frequencies. Some kind of filtering is therefore nec- essary to separate the two components. If no filtering is used, the colour information is interpreted as spurious fine details and the result is a grainy appearance over the entire picture. The question is what type of filter.
The cheapest solution is to use simple filters such as a bandpass or a notch filter that pass only the coarse and medium horizontal detail (lower about 3.5 or 3 MHz for PAL and NTSC, respectively) to the luminance cir- cuits and pass the bulk of the higher frequencies as colour information. For low-frequency luminance, the picture would suffer very little colour contamination or cross talk but at high frequency video where there are fine details such as striped clothing, noticeable artefacts in the form of ‘rainbow swirls’ appear in the picture at that location. Conversely, medium detail colour information cannot be used because there is too much lumi- nance contamination. Cross-talk from chroma C into luminance Y pro- duces waves of dots in the picture detail while cross-talk from Y into C results in spurious tints appearing as ‘rainbow swirls’ (Figure 14.3).
To avoid Y/C cross-talk, the two components must be separated as cleanly as possible. Hence, the use of comb filters.
You would recall from Chapter 2 that in order to avoid a dot pattern of alternate black and white dots caused by the insertion of the chrominance
information within the luminance bandwidth, the sub-carrier value was selected very carefully to ensure that the number of sub-carrier cycles per line ends with a half-cycle. This ensures that the sub-carrier clock is always 180° out-of-phase with the sub-carrier of the previous line in the same field. This fact is used by the comb filter to separate the two components Y and C. The technique involves the simple process of adding and subtracting the video contents of two successive lines. If line 1 in the odd field contains Y + C and line 3 contains Y – C, then by adding and subtracting we get
Similarly for the even field.
However, life is not as easy as that. The above assumes the Y and C components of two lines are the same. In practice, both Y and C change line-by-line as the picture content varies and along each line resulting in severe artefacts. For the effect on colour, consider a simple still picture, with a red square on top of a green square. In the middle of either square, consecutive horizontal scan lines are the same and the addition and sub-traction work adequately. However, where the squares meet, we pick up a pair of scan lines that differ profoundly, red versus green. What comes out in the picture is a yellow line. This is the result of two differing scan lines mixed together and yielding a third unrelated colour. More correctly, we see a yellow boundary two lines thick because the commingling of a red line and a green line happens twice per frame, first during the odd interlaced field and again during the even interlaced field. The yellow lines have a fine undulating dark–light pattern due to imperfect removal of the colour information from the luminance information. This is known as dot crawl. The undulations usually shift during the 25 or 30 frames per second refresh cycle. This artefact is entirely due to actions of the comb filter. Without a comb filter, there would be a perfectly sharp red to green horizontal boundary. To overcome these drawbacks, several techniques are available for TV manufacturers to use. There are four types of comb filters: 2L, 3L, 2D-3L and 3D.
The 2-line comb filter
This is the simplest and cheapest comb filter. It uses the simple add and subtract technique mentioned above. The Y and C components of consecutive lines in a frame are added and subtracted to produce luminance chrominance components, respectively. Compared with the notch filter, the 2-line (2L) comb filter improves Y/C cross-talk reducing the ‘rainbow swirls’ effect and improves horizontal resolution.
The 3-line comb filter
This technique uses three lines giving the first 50% weighting, the second 100% weighting and the third 50% weighting. If all three lines are identi- cal, the result is perfect since the two half strength lines are out-of-phase with the full strength middle line. Where the lines are not the same, the result is a vast improvement on the 2-line filtering. Let us go back to the picture of a red square above a green square. Starting from the top, we get perfect red mixtures until we get down to the boundary between the two squares. When we first encounter a green line we take it at half strength. The two preceding lines are a red line at full strength and another red line at half strength. In this 75% red–25% green mixture, the red overwhelms the green. As we move one line further down, we pick up the next green line at half strength, use the previous green line at full strength, and the line before that is a red line at half strength. This time the green is at 75% thus overwhelm- ing the red at 25%. Again this process happens twice, first in the odd field and then in the even field. So, the total number of finished scan lines derived from less than 100% red or less than 100% green in this example is four com- pared with the two in the 2L filter.
The 3-line (3L) comb filter provides better resolution than the 2L type, sharper horizontal boundaries and reduced ‘rainbow swirl’ effect.
The two-dimensional 3-line comb filter
This uses three lines in the same way as the 3L type. However, the weight- ing is not fixed. It is changed from left to right along the scan lines depend- ing on the similarity of the three lines. For example, where just the last two
of the three lines are the same, the comb filter mixes them 0, 50 and 50% to get near perfect filtering. Moving further along the line, the first two lines might be the same and the mixture is switched to 50, 50 and 0%. There may be intermediate mixtures too, used if the filter logic ‘was not sure’ whether the line above or the line below was a better match. Dot crawl can be eliminated over most of the picture and be made almost unnoticeable in the most difficult places. Crisp colour boundaries are achieved for most of the content of the picture. However, if neither the line above nor the line below matches, these comb filters cannot improve that part on the picture. An artefact will appear just as with the 2L comb filter. The 2D-3L type of comb filters are called two-dimensional (2D) because the mixture is varied vertically by using or not using the predecessor and/or successor lines and changed numerous times horizontally along the line.
The 3D (motion adaptive) comb filter
The 3D comb filter technique is qualitatively different. It uses the fact that the sub-carrier clock is always 180° out-of-phase with not just the sub-carrier of the previous line but of the previous frame as well. Thus, if line Lx in frame n has chroma C, then on the same line Lx on the previous frame n – 1 would have chroma –C; the chrominance component is phase reversed on adjacent frames.
The 3D comb filter uses adjacent frames (next and/or previous frames) to get a scan line that has the same content as the current line for correct Y/C separation. If there was no motion of the subject matter, the corre- sponding line in the next frame (625 PAL and 525 NTSC lines away) will have the same content as the line being processed. Its colour content, how- ever, is phase reversed. So, the sum of these lines is pure luminance and the difference is pure colour. Of course, if there is movement at that spot, the lines will differ and should not be commingled. The filter logic senses that, foregoes the ‘third dimension’ and go back to a method that 2D filters use. A good 3D comb filter contains within it a 2D comb filter. While there are elaborate processes for detecting motion, possibly involv- ing three frames instead of two, all that has to be done is sense whether portions of the lines are the same or different. This process is not perfect and is made more difficult if the picture is noisy (snowy). The circuits would also have to store an entire frame in a rolling data buffer since a line could not be drawn on the picture tube until the corresponding line in the next frame has arrived.
A typical 3D comb filter arrangement employed by Sony is illustrated in Figure 14.4. The composite video CBVS is fed into the colour decoder (IC1) from the video selector at pin 44. It comes out at pin 3 to go to IC2 (3D) and IC3 (3L) for comb filter processing. Two separate combinations of Y and C (YD and CD from IC2 and YL and CL from IC3) are made available for the 2-channel demux to select from. The selection is made by a signal from the
central control unit (CCU) to transistor Q1. A logic high signal will turn the transistor on and the demux will select the outputs from the 3D comb filter and vice versa. The CCU will assess the movement between successive frames and if there is movement, the outputs from the 3L filter will be selected, otherwise, that from the 3D filter will be chosen. The output from the colour decoder is in the form of Y, CR and CB (or YPrPb) which in this case is digitised by a analogue-to-digital converter (ADC) for further processing.