Skip to main content

Full text of "BBC Research and Development Report Number 1963-41"

See other formats


RESEARCH DEPARTMENT 

LINE- STORE STANDARDS CONVERSION: 
SUBJECTIVE EFFECT OF UNCONVERTED COMPONENTS 



Report No. T-113 
(1963/41) 



V.G. Devereux, M.A. (D. Maurice) 



This Report Is the property of the 
British Broadcasting Corporation and 
■ay not be reproduced in any form 
without the written permission of the 
Corporation. 



Report No. T-113 

LINE -STORE STANDARDS CONVERSION: 
SUBJECTIVE EFFECT OF UNCONVERTED COMPONENTS 



Section Title Page 



SUMMARY 1 



1. INTRODUCTION 1 



2. DETAILS OF EXPERIMENT . 1 

2.1. Description of Apparatus .... 1 

2.2. Conditions of Tests 2 

2.3. Experimental Procedure 3 



3. EXPERIMENTAL RESULTS ....... 4 



4. CONCLUSIONS .... ......... 6 



5. REFERENCES 7 



APPENDIX 



October 1963 Re P ort No - T" 113 

(1963/41) 

LINE -STORE STANDARDS CONVERSION: 
SUBJECTIVE EFFECT OF UNCONVERTED COMPONENTS 



SUMMARY 

This report describes experiments that were carried out in order to investi- 
gate the subjective effects of interference caused by 'break-through' of unconverted 
input signal to the output of a 625-line/405-line line-store standards converter. 
The effects of two different forms of 'break-through' were investigated; the first 
form was that which would result from an aperiodic coupling between the input and out- 
put of the converter and the second form was that resulting from a differentiating 
coupling. Experiments were carried out using pictures derived from slides, and curves 
have been plotted showing how the subjective effect of the two forms of interference 
varied with the amplitude of the interfering signal for three different types of 
subject matter. 

1. INTRODUCTION 

In a line-store standards converter 1 the redistribution of the video infor- 
mation on a new time scale is carried out by means of some form of 'redistributing' 
store. One form of such a store contains a pair of electronic switches and a storage 
device for each picture element in one line of the display. One of the problems of 
this method of conversion is that it is instrumentally difficult to avoid a certain 
amount of 'break-through' of the input signal to the output, resulting in an inter- 
ference pattern on a display of the output signal. The experiments described in this 
report were carried out in order to investigate the subjective effect of various levels 
of this interference and the results of these experiments help in specifying an 
acceptable limit to the spurious coupling which could be tolerated across the switches 
and storage devices in a converter. 

2. DETAILS OF EXPERIMENT 

2.1. Description of Apparatus 

A block diagram of the equipment used to simulate the effect of 'break- 
through' in a line-store standards converter is shown in Fig. 1. The two flying-spot 
scanners shown in this figure used identical slides; one scanner operated at the in- 



put standard of the simulated conversion while the other operated at the output stan- 
dard. In the tests, one scanner provided a 625-line video signal of 5 Mc/s band- 
width and the other scanner provided a 405-line video signal of 3 Mc/s bandwidth. 



input field 
drive pulses 



"output field 
drive pulses 



50c/s 



input 
pulse generator 



mains locking , " ' v ' I 
signal 




Fig. 1 - Block diagram of apparatus 



In line- store standards 
conversion, the field pulses of 
the input and output standards 
must be maintained precisely in 
phase with one another. In order 
to maintain this condition as 
rigidly as possible, the syn- 
chronizing- pulse generator of one 
scanner was controlled by a signal 
obtained from a phase- discriminator 
that compared the phases of the 
two trains of field-drive pulses. 
The residual jitter in the relative 
timing of these pulses had a 
magnitude of about 2 fis peak-to- 
peak, and a mean frequency of less 
than 1 c/s, this jitter being 
superimposed on a long term drift 
that was sufficiently slow to 
avoid increasing the visibility of 
the interfering signal. Tests 
showed that at levels near the 
threshold of visibility, the more 
rapid jitter had little effect on 
the results. 



The switch SI selected the type of 'break-through' to be simulated; the 
two types investigated were 'break- through' via (a) an aperiodic network and (b) a 
differentiating network, the latter network being the 'first derivative' path of an 
equalizer type TV/BQ/12. The amplitude of the interfering signal was controlled by a 
calibrated variable attenuator. The wanted signal plus interference was limited in 
bandwidth by a low-pass filter appropriate to the output line-standard, (3 Mc/s for 
405 lines, 5 Mc/s for 625 lines). 

After passing through the filter, the signal was finally displayed on a 
21-inch (53 cm) monitor synchronized by pulses from the output- standard pulse generator. 
Neither the wanted nor the unwanted video signals were accompanied by synchronizing 
pulses, since these would almost certainly be absent in the switch and store sections 
of a converter. 

2.2. Conditions of Tests 



The monitor was adjusted to give a brightness of 20 ft-L (215 asb) for a 
signal input corresponding to white and the ambient lighting was such that the monitor 
screen reflected about 0°25 ft-L (2°7 asb) when the cathode-ray tube beam current was 
cut off. 



Tests were carried 
out using three different 
slides; these are shown in 
Fig. 2, The slide of the 
grey-white step was constructed 
in order to give the greatest 
possible visibility of inter- 
ference at a given amplitude 
of interfering signal. The 
grey left-hand half of the 
wanted signal display obtained 
from this slide was adjusted 
to a brightness of 1 ft-L, 
(10'7 asb) so as to be close 
to that brightness at which an 
added interfering signal giving 
a stationary pattern would be 
most visible under the given 
viewing conditions (see 
Appendix) . A signal corres- 
ponding to white was used for 
the right-hand half of this 
display in order to provide 
the maximum amplitude of inter- 
fering signal. 

Slides 2 and 3 were 
chosen to show the difference 
in the subjective effect of 
the interference in two types 
of scene; slide 2 contained a 
large proportion of plain area, 
while slide 3 contained much 
detailed information. 

2.3. Experimental Procedure 

With the switch Si 
in the position that by-passed 
the differentiating network, 
various levels of the input 
signal were added to the output 
signal.. The subjective effect 
of the resulting interference 
on the display was assessed by 
six observers who were seated 
at between five and seven 
times the picture height from 
the monitor. These observers 
were asked to express their 
results according to the scale 




Slide 1 - Grey-white step 





'.'■■■. 

wmwaMm 

Slide 2 - Girl with fan 






-Kfe:. 



'%■■■■■ ,(:)::■ ■■■? : %,:, ■'■■?% ■'. 



^§ 



.?''.■■■ ;i^''' / - 



U'i^V* ■'.'$<! u:-: 




"'■■.,:■;.:■::,"■'' r: i2«-$ -si-: 








,« ;t: s, .;,;■'< - 


■&> ?.:-% ~:T' 


— ^ 


::,<■**..■:? 




Slide 3 - V.I. P. studio 
Fig. 2 - Slides used in tests 



given in Table 1 for each of the three slides in turn. The differentiating network 
was then connected in the circuit and the tests were repeated. 

This procedure was followed first using the 405- line signal as interference 
and the 625- line signal as the wanted picture, and secondly with 625 lines as the 
interfering standard and 405 lines as the wanted standard. 

TABLE 1 
Scale of Subjective Effect of Interference 



Subjective Effect of Interference 


Score 


Imperceptible 


1 


Just perceptible 


2 


Definitely perceptible but not disturbing 


3 


Somewhat objectionable 


4 


Definitely objectionable 


5 


Unusable 


6 



3. EXPERIMENTAL RESULTS 

The results of the tests are shown in Figs. 3 and 4 in which the average 
marking of the six observers has been plotted against the attenuation, in decibels, 
of the interfering signal relative to the wanted output signal. 

Since the attenuation of a differentiating network is dependent on frequency, 
being given by: 

Attenuation (dB) = c - 20 logi / 

where / = frequency 

and c = constant, 



the horizontal axis in Fig. 4 has been calibrated in terms of the network attenuation 
at a particular frequency, i.e. 3 Mc/s. The horizontal axes of both Figs. 3 and 4 
are calibrated in terms of the attenuation provided by unwanted coupling between the 
input and output of a converter, relative to the attenuation provided by the wanted 
coupling. In Fig. 3, a value of dB corresponds to wanted and unwanted couplings 
having equal attenuations; in Fig. 4, a value of dB corresponds to an unwanted 
coupling that provides an attenuation at 3 Mc/s which is equal to the attenuation 
provided by the wanted (aperiodic) coupling. 

Considering Fig. 3, the results show that, when the interfering signal was 
coupled to the wanted signal via an aperiodic network, it was necessary to attenuate 
the interfering signal by at least 47 dB relative to the wanted signal in order that 



£-8 



>° 



:*s 2 











































I 
\ 




















\ 






















\ 




















\ 






















\ 


A 














\ 






















\ 




V 
















\ 
1 


\. 




S 


^-« ' 
















\ 






s ^ 














N 




-x- 




^ 


te^o^ 





25 30 35 40 



45 50 




30 35 40 45 



attenuation of interfering signal relative to wanted signal, dB 
a b 

Fig. 3 - Visibility of interference when interfering signal added 

via an aperiodic network 

a 405-line interfering signal added to 625-line wanted signal 
6 625-line interfering signal added to 405-line wanted signal 

o Slide 1 - Grey-white step 

•• Slide 2 - Girl with fan 

* Slide 3 - V.I. P. studio 



5 


" 


































ST 
























































k 


















V 
\ 
















>> O 




\ N 1 


\ 












- °2 




\ 


\ 
















' 




\ 


















V 

\ 


V 




■ / . 


\ 












NU 


. --x"" 




^~""^l 


1 





b 


' 


































4 




































\ 










































































1 


^\ 




















X \ 


~A 






















.'* 


'n. 


1 
















-**"^ - 



10 15 20 5 10 

attenuation of 3Mc/s component of interfering signal, dB 

a b 



15 



20 



Fig. 



4 - Visibility of interference when interfering signal added 

via a differentiating network 

a 405-line interfering signal added to 625- line wanted signal 
625-line interfering signal added to 405-line wanted signal 

o Slide 1 - Grey- white step 

♦ Slide 2 - Girl with fan 

« Slide 3 - V.I. P. studio 



it be imperceptible on the most critical scene used in the tests; further, the 
attenuation required to produce a given visibility of interference, using the most 
critical slide, was about 10 dB greater than that required with the least critical 
slide. The reason for a given amount of interference being least visible on slide (3) 
was probably due to the fact that this slide contained much detailed information and 
few large plain areas, with the result that the interference tended to be masked by 
the wanted information. 

The results plotted in Fig. 4 show that, when the interfering signal was 
coupled to the wanted signal through a differentiating network, it was necessary to 
attenuate the 3 Mc/s component of the interfering signal by at least 16 dB in order to 
render the interference imperceptible when using the most critical slide. The fact 
that several of the curves in Fig. 4 show a minor peak between 14 dB and 20 dB is 
considered to be of no significance and was probably caused by observers confusing 
fine detail in the wanted picture with the interference when the latter had a very low 
level. It can also be seen that, with a differentiating coupling, the attenuation 
required in order to produce a given visibility of interference was less dependent on 
the type of slide display than when the coupling was aperiodic, especially when using 
a 405- line display. This effect was probably due to the fact that the masking effect 
of detailed information in the wanted picture was counteracted by an increase in the 
level of interference caused by high-frequency video signals arising from detailed 
information. 

Finally, both Figs. 3 and 4 show that interchanging the line standards of 
the wanted and interfering signals made little difference to the visibility of the 
interference for a given attenuation in the coupling network. 



4. CONCLUSIONS 

In a 625/405 line- store standards converter, any spurious coupling of an 
aperiodic nature between the input and output should have an attenuation of at least 
47 dB, relative to the wanted coupling, if the resulting interference is to be imper- 
ceptible. With an attenuation of 40 dB, the interference will be just perceptible on 
the most critical scenes. 

If the input is spuriously coupled to the output through a differentiating 
network then, taking the amplitude of the 3 Mc/s component of the input signal as a 
reference level, the attenuation provided by the spurious coupling at 3 Mc/s relative 
to that provided by the wanted (aperiodic) coupling must be at least 16 dB in order 
that the interference on the display of the output signal should be imperceptible. 
This figure of 16 dB at 3 Mc/s corresponds to an attenuation at a frequency / Mc/s 
given by: 

Attenuation (dB) = 16 + 20 log 10 (3//) dB. 

where the video frequency / is expressed in Mc/s. 

An attenuation of 10 dB allows the interference to be just perceptible on the most 
critical scenes. Interchanging the standards of the input and output signals makes, 
in general, less than 1 dB difference to the attenuation required for a given visi- 
bility of interference. 



The above results were obtained using still pictures only; with moving 
pictures the attenuation of the unwanted signal required for a given visibility of 
interference can be somewhat higher 2 since there would be relative movement between 
the interference and the wanted picture detail. The attenuation increase required by 
moving scenes would probably be about 6 dB when the interference is definitely visible, 
but the attenuation required for interference at the threshold of perception would 
probably be about the same for moving pictures as still pictures. 



REFERENCES 

1. 'An Outline of Synchronous Standards Conversion Using a Delay-Line Inter- 
polator' , Research Department Report No. T-096, Serial No. 1962/31. 

2. Schade, Otto H. , 'Optical and Photoelectric Analog of the Eye', J. Opt. 
Soc. Am., September 1956, Vol. 46, No. 9. 

3. 'Line-Store Standards Conversion: Subjective Effect of Switch and Store 
Tolerance' , Research Department report in preparation. 

4. Moon, P., and Spencer, D.E., 'The Visual Effect of Non-Uniform Surrounds', 
J. Opt. Soc. Am., March 1945, Vol. 35, No. 3. 

5. 'The Variation in the Visibility of Interference Over the Grey Scale of a 
Television Picture' , Research Department Report No. T-092, Serial No. 
1962/27. 



APPENDIX 

Relative Visibility overtheGrey Scale on a Television Display 
of an Interfering Signal added to the Displayed Signal 



This Appendix describes a method of calculating the variation in the visi- 
bility of an interfering signal over the grey scale of a television display. The 
results obtained show how the visibility of interference is affected by the following 
factors: 

(a) The gamma of the display tube transfer characteristic. 

(b) The luminance of an unexcited area of the display caused by ambient 
illumination and flare in the tube. 

(c) The adaptation condition of the eye viewing the display. 

The calculations are based on two assumptions, these being as follows: 

(i) It is assumed that the luminance B of a display tube is related to the 
applied signal voltage E by the equation: 



B - Bo = AE7 



(1) 



whe 



B = luminance of an unexcited area of the display caused by ambient 
illumination and flare in the tube. 



and 



20 r 



A and y are constants. 



10 



£ 3 

3> 2 
i 

CD 



i&5 



0-3- 



0-2- 



OT 







- 


~~~ 







~ 






-■■■ 






































, _ .... / 






















\t - 


2 


42 
















/ : i 


i ' j I ■■ 


,. ! ! :/ ! i _L_ . 




/ ; 








! M 



10 10 2 

applied signal voltage as percentage of 
peak white signal 

Fig. 5 - Luminance (B - B ) plottea 
against applied signal voltage 



The errors involved in this assumption 
are quite small for the display tube used in the 
experiments described in this report; Fig. 5 
shows a curve of luminance plotted against 
applied signal voltage which was obtained from a 
similar type of display tube. 

If the maximum luminance 6 results 
from an applied signal of amplitude E , then: 

A = (B M - B )/E y M (2) 

Therefore, from equations (1) and (2): 



(B - B /B M - B Q ) = (E/E M )' 



(3) 



By differentiating equation (3), it 
can be seen that the difference AB in the 
luminances of two areas on the display, caused 
by a small difference AE in the applied signal 
voltages, may be conveniently expressed as: 

AB ^y{B u - B /B - B ) l/7 . (B - B ) . AE/£, (4) 



(ii) It is assumed that, under the adaptation conditions of the eye existing when 
an average television display is being viewed, the difference AS in the sensations 
caused by light from two adjacent areas differing in luminance by a small amount AB 
is given by: 



where 



AS = k AB/(B + B t ) 
B = luminance of the brighter of the two areas 



(5) 



Bi - constant for a given adaptation condition of the eye 

k - constant for two areas of a given shape subtending a 
given angle at the fovea of the eye. 



Equation (5) is an approximation to a 
formula given by Moon and Spencer; 4 their data 
were obtained from experiments carried out in 
order to determine the threshold of perception of 
the difference in luminance between a circular 
object field and a surrounding annular test field, 
these fields subtending angles of 1° and 1%° at 
the fovea of the eye respectively. (See Fig. 6.) 

In Fig. 7, a comparison has been made 
between curves obtained from the Moon and Spencer 
formula and from equation (5). These curveis 
show the variation in the Fechner fraction with 
luminance for three different adaptation condi- 
tions of the eye. 



surrounding field at luminance B s 




test field 



object field 



Fig. 6 - Fields used in Moon and 
Spencer experiments 







001 l 



0O01 



0O1 



0-1 10 

luminance of test field B, ft - 



10 



100 



Fig. 7 - Variation of Fechner fraction with luminance 

of test field 
~ Curves given by Moon and Spencer formula for threshold of perception 
— Curves given by As = k Ab/CB+Bj) plotted for As/k = 0-017 



10 

The Moon and Spencer curves have been plotted for the adaptation conditions 
existing when the whole of the field surrounding the test field is at a constant 
luminance, denoted by B s in Fig. 7. The values of AS/& and J?i in equation (5) were 
chosen so that the curves obtained fitted the Moon and Spencer curves as closely as 
possible. It can be seen that there is close agreement between the two sets of 
curves except when the surrounding luminance is very low. 

It is of interest to note that if equation (5) were a true representation 
of the relation between luminance and sensation, then departures of the Fechner 
fraction AB/B from a constant value could be explained by assuming that a veiling 
glare of luminance Bx exists in the eye. This assumption would mean that the lumi- 
nance perceived by the eye would be B + Bi and hence for a given difference in 
sensations AS, : 

ABe/Be = C 

where Be (the luminance perceived by the eye) - B + B^ 

and C is a constant. 

From equations (4) and (5), the difference in the sensations caused by the 
light from two adjacent areas on a television display which differ in brightness as 
the result of an interfering signal of amplitude AB, is given by: 

AS = ky (B H - B /B - B ) l/T . (B - B /B + BJ . A?/£„ (6) 

For a given value of AE, AS reaches a maximum value at a brightness given by: 

B = 7fl + (7 - 1) B t (7) 

Suitable values for B Q , B ± and y applicable to the tests described in the 
report are: 

B = 0-25 ft-L (2-7 asb) 

Bj. = 0*3 ft-L (3-2 asb) 

y = 2*4 

Inserting these values in equation (7) gives 1*02 ft-L (10*9 asb) as the 
brightness at which AS reaches a maximum. 

A curve obtained from equation (6) is shown in Fig. 8. This figure shows 
the attenuation (dB) of an added interfering signal which is required to maintain the 
visibility of the interference (i.e., AS) constant plotted against the luminance of 
the area of the display on which the interference is superimposed. 

Also given on this figure is an experimental curve obtained by Geddes. 
This curve was obtained from subjective tests on the visibility of stationary inter- 
fering patterns on a television display; the maximum luminance of the display was 
20 ft-L (215 asb) while the luminance of unexcited areas was 0*25 ft-L (2*7 asb). 



11 



10 

a q 






















































erfering signal, ( 
♦ constant] 

)1 0) vl Q) 


























































\ 






























,' / 








































































\ ] 


I 












































'" < A 








\ 
\ 














































level of 
20 log 








\ 




















J 
































\\ 


\ 












































tive 
[i.e. 

-> 










N 


■ 










































o 

i- 

-1 














> 
























































































1 



















to 2 
















10 J 
















o 4 



luminance of display fi, f t - L 



Fig. 8 - Relative amplitude of interfering signal to produce constant 
visibility of interference plotted against luminance of display 



Carve obtained from equation /SE/E U 



&S/ky.(B + BjB ■ B Q ) . (B 



V B JK 



*0> 



i/y 



plotted for B Q = 0-25 ft-L (2-7 asb) , B ± = 0-3 ft-L (3-2 asb), y = 2-4 



Experimental curve obtained by Geddes. 



B = 0-25 ft-L (2-7 asb) 



Differences between the theoretical and experimental curves are to be 
expected since certain approximations have been made in the theoretical analysis, 
the most important of which are: 

(a) It has been assumed that the transfer characteristic of a display tube can 
be expressed in terms of a constant value of y equal to 2*4, whereas the 
experimental results are related to the transfer characteristic of an 
actual display tube. 

(b) As mentioned above, equation (5) relating sensation and brightness is not 
entirely correct. In addition, the value of B x used in this equation is 
somewhat arbitrary and varies according to the experimental conditions. 

Equation (6) can also be used to predict, quantitatively, the ratio A£/£ of 
interfering to wanted signal at which the interference appearing on a display corres- 
ponds to the threshold of perception. For a value of AS/fe = 0*017, equation (5) was 
found to agree substantially with the curve representing the Moon and Spencer data 
corresponding to the threshold of perception (see Fig. 7). (It should be noted that 
this value of AS/fe only applies to experimental conditions similar to those used by 
Moon and Spencer; with test fields subtending smaller angles at the fovea of the eye 
AS/fc would be greater.) In the tests described in this report, the peak value E of 
the wanted signal corresponded to a luminance B of 20 ft-L (215 asb). Substituting 
the aforementioned values of B , B it B , AS/fe and y in equation (6) gives: 

20 log &E/E M = -50 dB 

when B = 1-02 ft-L (10-9 asb) 

This figure compares quite well with the experimental figure of -47 dB 
which was obtained for the 'Grey-White Step' display (see Fig. 3). 



BRH 



Printed by B.B.C. Research Department, Eingswood Warren, Tadworth, Surrey