1、EIA TEB27 BB m 3234600 0007253 9 m TEPAC ENGINEERING B U LL E T IN Relating Display Resolution And Addressability TEB27 AUGUST 1988 ELECTRONIC INDUSTRIES ASSOCIATION ENGINEERING DEPARTMENT EIA TEB27 88 = 3234b00 0007254 O i NOTICE EIA Engineering Standards and Publications are designed to serve the
2、public interest through eliminating misunderstandings between manufacturers and purchasers, facilitating interchangeability and improvement of products, and assisting the purchaser in selecting and obtaining with minimum delay the proper product for his particular need. Existence of such Standards a
3、nd Pub- lications shall not in any respect preclude any member or non-member of EIA from manufacturing or selling products not conforming to such Standards and Publications, nor shall the existence of such Standards and Publications preclude their voluntary use by those other than EIA members, wheth
4、er the standard is to be used either domestically or internationally. Recommended Standards and Publications are adopted by EIA without regard to whether or not their adoption may involve patents on articles, materials? or processes. By such action, EIA does not assume any liability to any patent ow
5、ner, nor does it assume any obligation whatever to parties adopting the Recom- mended Standard or Publication. This TEPAC Engineering Bulletin was developed by the JT-20 Committee on Cathode-Ray Display Devices. Pub1 ished by ELECTRONIC INDUSTRIES ASSOCIATION Engineering Department 2001 Eye Street,
6、N.W. Washington, D.C. 20006 I- PRICE: $10.00 Published in U.S.A. i a EIA TEE27 88 W 3234600 0007255 2 9 TEB27 Page 1 RELATING DISPLAY RESOLUTION AND ADDRESSABILITY Gerald M. Murch and Robert J. Beaton imaging Research Lab, Tektronix Laboratories, Tektronix, Inc., ABSTRACT The perceived image quality
7、 of a digital display is affected by two independent system characteristics, namely resolution and addressability, This paper discusses a quantitative procedure for selecting optimal levels of resolution and addressability based upon the performance constraints of human vision. INTRODUCTION A great
8、deal of confusion exists about the precise meaning of the terms Resolution and Addressability as applied to display systems, Additionally, once clarity of meaning is achieved, the problem remains of designing a system in which these two quantities are properly related to each other to maximize the q
9、uality of the displayed image. Such a relationship is one in which the image meets some perceptual cfiterion of image quality. This paper, then, is concerned with clarifying the meaning and specification of resolution and addressability in Cathode-Ray Tube (CRT) based display systems arid with prese
10、nt- ing a visually derived metric for selection of the optimal relationship of resolution and addressability . In essence, resolution is a property of the design of the display device. It is derived from the width of a line or spot imaged on the screen: The narrower the line or the smaller the spot,
11、 the higher the resolution. From the measured line width, resolution can be specified in a number of ways, such as lines per unit distance, Modulation Transfer Function (MTF), spot size, etc. We will consider the line width at 50% of the maximum lumi- nance intensity, since simple conversions exist
12、to translate the various metrics of resolu- tion 1,2. Addressability is a characteristic of the display controller and represents the ability to select and activate a specific point of x,y coordinate on the screen. On rastered displays, this is usually stated in terms of the number of lines scanned
13、from the top to the bottom of the display screen as well as the number of points along each raster line. Since addressability is controiled by the hardware driving the CRT, and since resolution is determined by the design of the CRT, these two display characteristics are independent . EIA TEB27 88 W
14、 3234600 0007256 4 7 TEB2 7 Page 2 of one another, However, to obtain high levels of image quality, certain relations need to be maintained between resolution and addressability. For example, if resolution is too low (large spot sizes), successive lines will over-write preceding lines. Under some co
15、n- ditions, this may produce image artifacts such as false contours. Conversely, if addressa- bility is too low (large spot separations), then adjacent raster lines will not merge and they will appear as visible stripes. RELATING RESOLUTION AND ADDRESSABILITY We assert that a primary goal in enginee
16、ring a visual display system is to attain sufficient image quality to maximize the transfer of “information“ from the display screen to the human operator. Although numerous factors contribute to overall image quality (e.g., ambient illumination, screen format, etc.), resolution and addressability d
17、irectly impact two fundamental criteria underlying this design goal. The est criterion, which we have termed the adjucent raster line (pixel) requirement, states that the raster structure of a display must be imperceptible to an operator located at a typical (46 cm) viewing distance. This requiremen
18、t is intended to eliminate visible “noise,“ which arises from the discrete picture elements of digital display systems, and which bears no relevant information for the operator. Display systems that meet the adjacent raster line (pixel) criterion present uniformly bright solid-lled areas and alphanu
19、meric characters, which appear continuously constructed and highly legible. The second image-quality criterion, termed the alternate raster line (pixel) requirement, states that individual lines (pixels) within an alternating on-off-on-off pattern must be visible to an operator from a typical viewin
20、g distance. This requirement optimizes the visibility of high spatial frequency components, such as narrow lines and fine details within an image. For a CRT system with a smoothly decreasing MW, optimizing the alternate raster line (pixel) criterion also optimizes the information transfer of low spa
21、tial fkequency components as well 3. The two above-mentioned image-quality criteria place opposing demands upon the optimal specification of display resolution and addressability. For example, increases in display ddressability favor the adjacent raster line (pixel) criterion since the modulation (l
22、uminance contrast) between adjoining raster lines is reduced; however, this same reduc- tion in modulation also reduces the detectability of individual lines within an on-off-on- off pattern, thereby disfavoring the alternate raster line (pixel) criterion. A similar trade- off occurs with changes in
23、 display resolution. RESOLUTION/ADDRESSABILITY RATIO In order to asses whether or nt a display system satisfies the two image-quality criteria mentioned above, the system designer must determine the modulation between adjacent and alternate raster lines (pixels). Then, these modulation values must b
24、e evaluated in terms of human visual sensitivities to modulation at specific spatial frequencies. It is desired that the modulation between adjacent raster lines (pixels) be below a minimum . EIA TEB27 B m 3234600 0007257 b m TEB2 7 Page 3 level required for visual stimulation, while the modulation
25、between alternate raster lines (pixels) should exceed this minimum visual stimulation level. The evaluate display systems over a wide range of resolution and addressability, we employ a simple metric, termed the Resolution/Addressability Ratio (RAR), given as W RAR=- S in which W denotes the full-wi
26、dth of a raster line (pixel) profile at one-haif its maximum luminance intensity and S denotes the peak-to-peak separation between adjacent raster lines (pixels). For example, a 48-cm diagonal display with a height of 27.5cm and an addressability of 1024 lines has a peak-to-peak separation of 27.5/1
27、024 = 0.27 mne. Assuming a 0.38-mm wide spot profile, the resulting RAR value is 1.41. By using a numerical simulation program, modulation values for adjacent and alternate raster lines (pixels) were determind under various combinations of resolution and addressability and, hence, RAR values. For ea
28、ch RAR value, the simulation program con- structed a waveform that represented the luminance pattern across several raster lines (pixels). These calculations were performed by convolving a Gaussian spot profile hav- ing a specific width (W) with a series of unit delta functions spaced at a specic se
29、para- tion (S), The modulation values, determined by Fourier analysis of the resulting periodic raster (pixel) structure, are shown in Figure 1 by the curve labeled “Adjacent Pixel.“ The observed trend in modulation, over a range of RAR values from 0.35 to 2.4, can be described by (2) in which M den
30、otes modulation and RAR is defined by Eq. 1. As an example, for a display with a resolution of 0.38 mrn and addressability of 480 lines within a 27,5-cm vertical display area, the modulation between adjacent raster lines (pixels) is 0.43. 2 M =- exp 3.6(RAR )-7 .O(RAR )2 +(RAR )3, IC With Eqs. 1 and
31、 2, the alternate raster line (pixel) modulation can be computed easily by doubling the separation value used in the corresponding adjacent raster line (pixel) calcu- lation; that is, the RAR value for the alternate criterion is equal to one-half of the RAR value for the adjacent criterion. Example
32、calculations are shown in Figure 1 by the curve labeled “Alternate Pixel.“ /y- EIA TEE27 88 m 3234600 0007258 8 m 1 .o I 0.8 - m - 0.6 - 0.4 - - 0.2 - - 0.0 - I TEB27 Page 4 Modulation Resolution /Addressability Ratio Figure 1. Adjacent and alternate raster line (pixel) modulation as a function of R
33、AR. SELECTING AN OPTIMAL RESOLUTION AND ADDRESSABILITY The final step in the evaluation procedure is to assess the detectability of adjacent and alternate raster line (pixel) modulation by a human operator. For this purpose, a classic nding from the field of visual science known as the Contrast Sens
34、itivity Function (CSF) is used, which describes the minimum modulation needed by the visual system to detect a sine-wave pattern of a given spatial frequency. Figure 2 present the 90% population CSF for she-wave patterns subtending at least 5 degrees of visual angle and having an average luminance o
35、f 10 cdh2 4. This CSF curve is described by the following least- squares regression equation: M=bo expbl(u) + Ma2) + bdW4)1, (3) in which M denotes the modulation required for detection at spatial frequency u), expressed in cycles per degree of visual angle. Values for the regression coefficients bo
36、, bl, b2, and b3 are 1,7062 x -2.3161 x loe3, and 0.2000 x lo4, respec tive1 y. 201.6188 x -7 Modulation EIA TEB27 88 m 3234600 0007259 T m 1 .o 1 I I I I I TEB2 7 Page 5 Spatial frequency (cycles /degrees) Figure 2. Human contrast sensitivity function with 90% population limits (broken lines). Equa
37、tion 3 can be used in conjunction with Eq. 2 to determine the visibility of adjacent and alternate raster line modulation. Before proceeding, however, it is necessary to determine the spatial frequency, in cycles per degree of visual angle, corresponding to the adjacent and alternate raster lines (p
38、ixels) and to rewrite Eq. 3 in a form that can be used directly with Eq. 2. For each image-quality criterion, spatial frequency is related to the separation between the raster lines (pixels) under consideration, since the peak-to-peak separation equals one period of the waveform cycle. Therefore, us
39、ing the fact that S = W + RAR (see Eq.l), a separation value can be converted into cycles per degree of visual angle by ri 1 in which D denotes the viewing distance of the operator from the display screen and S, W, and RAR are defined by Eq. 1. The right-hand side of Eq.4 can be substituted into Eq.
40、 3 to express the CSF as a function of RAR for various levels of display resolution. Figure 3 presents an example of the display evaluation procedure, where a resolution level was chosen and an optimal display addressability was desired. By the use of Eq. 2, the adjacent and alternate raster line (p
41、ixel) modulations were computed for various , addressabilities and, therefore, various RAR values, as shown by the two decelerating curves. Next, Eqs. 3 and 4 were used to determine the CSF values for the adjacent and alternate raster line (pixel) criteria, as shown by the two accelerating curves. O
42、ne bound on the optimal addressability is provided by the intersection of the adjacent raster line (pixel) modulation curve with the corresponding CSF curve, while another bound is pro- vided by the intersection between the two curves for the alternate raster line (pixel) cri- terion, EIA TEB27 dd m
43、 3234b00 00072b0 b m x TEB2 7 Page 6 l.OI I Resolution I I = 0.254 mm I i Modulation 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 Resolution Addressability Ratio Figure 3. Modulation and visual contrast sensitivity values for the adjacent (squares) and alternate (triangles) raster line (pixels) c
44、riteria. For a selected resolution level, a range of addressabilities will satisfy the adjacent and alternate raster line (pixel) criteria, In these situations, it will be convenient to choose the adjacent RAR limit since the resulting addressability maximizes the visibility of high spa- tial freque
45、ncy components while maintaining the raster modulation at a “just detectable“ level. Table 1 lists the RAR limits for a variety of frequently used resolution levels. Note that the alternate raster line (pixel) limit is twice as large as the corresponding adjacent raster line (pixel) limit. Table 1.
46、Adjacent and Alternate Raster Line (Pixel) Limits. RESOLUTION - ITII ADJACENT PIXEL LIMIT ALTERNATE PIXEL LIMIT O. 127 0.7 1 1.42 0.254 0.92 1.84 0.38 1 1 .o2 2.04 0.508 1 ,o9 2.18 0.635 1.12 2.24 0.762 1.18 2.36 CONCLUSIONS Resolution and addressability play key roles in determining the image quali
47、ty of digital- display systems. In this paper, we have discussed a straightforward procedure to match these display characteristics to human visual capacity. EIA TEE27 88 U 3234b00 00072bL 8 W REFERENCES TEB2 7 Page 7 Shen, S., Electronic Displays (John Wiley & Sons, Inc. New York, 1980). Keller, P.
48、A., “A Survey of Data-Display CRT Resolution Measurement Tech- niques,“ in Society for Information Display, Seminar Lecture Notes, (Los Angeles, California: Society for Information Display 1984) pp. 2.2a- 1-2.2a-28. Beaton, R.J., A Human-Performance Based Evaluation of Quality Metrics for Hard-Copy
49、and Soft-Copy Digital Imaging Systems, (Unpublished doctoral disser- tation, Virginia Polytechnic Institute and State University, Blacksburg, Virginia, 1984). Snyder, H.L., “Human Visual Performance and Flat Panel Display Image Quality,“ Technical Report No. HFL-80-1 IONR-80-1, (Blacksburg, Virginia Polytechnic Institute and State University, Human Factors Laboratory, 1980). o
