CHAPTER 25 ILLUMINATION

Transcription

1 CHAPTER 25 ILLUMINATION Peter R. Boyce Consultant Canterbury, Great Britain 1 INTRODUCTION MEASUREMENT OF ILLUMINATION Photometric Quantities Colorimetric Quantities Instrumentation PRODUCTION OF ILLUMINATION Daylight, Sunlight, and Skylight Electric Light Sources Control of Light Distribution Control of Light Output FUNCTIONAL CHARACTERISTICS OF THE HUMAN VISUAL SYSTEM Visual System Structure Wavelength Sensitivity Adaptation Color Vision Receptive Field Size and Eccentricity Meaningful Stimulus Parameters EFFECTS ON THRESHOLD VISUAL PERFORMANCE Visual Acuity Contrast Sensitivity Function Temporal Sensitivity Function Color Discrimination Interactions Approaches to Improving Threshold Visual Performance EFFECTS ON SUPRATHRESHOLD VISUAL PERFORMANCE Relative Visual Performance Model for On-Axis Detection Visual Search Visual Performance, Task Performance, and Productivity Approaches to Improving Suprathreshold Visual Performance EFFECTS ON COMFORT Symptoms and Causes of Visual Discomfort Lighting Conditions That Can Cause Discomfort Comfort, Performance, and Expectations Approaches to Improving Visual Comfort INDIVIDUAL DIFFERENCES Changes with Age Helping People with Partial Sight Consequences of Defective Color Vision OTHER EFFECTS OF LIGHT ENTERING THE EYE Circadian System Positive and Negative Affect TISSUE DAMAGE Mechanisms for Damage to the Eye and Skin Acute and Chronic Damage to the Eye and Skin Damage Potential of Various Light Sources Approaches to Limiting Damage EPILOGUE 666 REFERENCES INTRODUCTION Illumination is the act of placing light on an object. By providing illumination, stimuli for the human visual system are produced and the sense of sight is allowed to function. With light, we can see; without light, we cannot see. This chapter is devoted to describing how to measure and produce illumination, the effects of different lighting conditions on visual performance and visual comfort, the photobiological and psychological effects of illumination and the risks inherent in exposure to light. 643 Handbook of Human Factors and Ergonomics, Third Edition. Edited by Gavriel Salvendy Copyright # 2006 John Wiley & Sons, Inc.

2 644 EQUIPMENT, WORKPLACE, AND ENVIRONMENTAL DESIGN 2 MEASUREMENT OF ILLUMINATION 2.1 Photometric Quantities Light is a part of the electromagnetic spectrum lying between the wavelength limits 380 to 780 nm. What separates this wavelength region from the rest is that radiation in this region is absorbed by the photoreceptors of the human visual system, which initiates the process of seeing. The most fundamental measure of the electromagnetic radiation emitted by a source is its radiant flux. This is a measure of the rate of flow of energy emitted and is measured in watts. The most fundamental quantity used to measure light is luminous flux. Luminous flux is radiant flux multiplied by the relative spectral sensitivity of the human visual system over the wavelength range 380 to 780 nm. The relative spectral sensitivity of the human visual system is based on the perception of brightness associated with each wavelength. In fact, there are two different relative spectral sensitivities, sanctified by international agreement arranged through the Commission Internationale de l Eclairage (CIE, 1983, 1990). There are two relative spectral sensitivities because the human visual system has two classes of photoreceptor: cones, which operate primarily when light is plentiful, and rods, which operate when light is very limited. These two photoreceptor types have different spectral sensitivities: the day photoreceptor, the cones, being characterized by the CIE standard photopic observer, and the night photoreceptor, the rods, being characterized by the CIE standard scotopic observer (Figure 1). Luminous flux is used to quantify the total light output of a light source in all directions. Although this is important, for lighting practice it is also important to be able to quantify the luminous flux emitted in a given direction. The measure that quantifies this concept is luminous intensity. Luminous intensity is the luminous flux emitted per unit solid angle, in a Relative luminous efficiency b c a Wavelength (nm) Figure 1 Relative luminous efficiency functions for (curve a) the CIE standard photopic observer and (curve b) the CIE standard scotopic observer. The CIE standard photopic observer is based on a 2 field of view. Also shown (curve c) is the relative luminous efficiency function for a 10 field of view in photopic conditions. specified direction. The unit of measurement is the candela, which is equivalent to a lumen per steradian. Luminous intensity is used to quantify the distribution of light from a luminaire. Both luminous flux and luminous intensity have associated area measures. The luminous flux falling on a unit area of surface is called the illuminance. The unit of measurement of illuminance is the lumen per square meter or lux. The luminous intensity emitted per unit projected area in a given direction is the luminance. The unit of measurement of luminance is the candela per square meter. The illuminance incident on a surface is the most widely used electric lighting design criterion. The luminance of a surface is a correlate of its brightness. Table 1 summarizes these photometric quantities and the relationship between illuminance and luminance. Unfortunately for consistency, photometry has a long history that has generated a number of different units of measurement for illuminance and luminance. Table 2 lists some of the alternative units, together with the multiplying factors necessary to convert from the alternative unit to the Systém International (SI) units of lumens per square meter for illuminance and candela per square meter for luminance. SI units will be used throughout this chapter. Table 3 shows some illuminances and luminances typical of commonly occurring situations. 2.2 Colorimetric Quantities The photometric quantities described above do not take into account the wavelength combination (i.e., the color) of the light being measured. There are two approaches to characterizing color, the color atlas and the CIE colorimetry system Color Atlases The color atlas, as its name implies, is a physical, three-dimensional representation of color space. It is three-dimensional because colors have three separate subjective attributes: hue, brightness, and strength. Hue tells us whether the color is primarily red or yellow or green or blue. Brightness tells us to what extent the color transmits or reflects light. Strength tells us whether the color is strong or weak. Several different color atlas systems are used in different parts of the world (Wyszecki and Stiles, 1982). Probably the most widely used atlas is the Munsell Book of Color available from the Munsell Color Company. Figure 2 shows the three-dimensional color space of the Munsell atlas. The position of any color is identified by an alphanumeric code made up of three terms: hue, value, and chroma (e.g., a strong red is given the alphanumeric 7.5R/4/12). Hue, value, and chroma are related to the three attributes of color: hue, brightness, and strength, respectively. Building materials, such as paints, plastic, and ceramics, are commonly classified in terms of a color atlas CIE Colorimetric System Sometimes, it is necessary to quantify the color of a light or a surface before either exists. To meet this

3 ILLUMINATION 645 Table 1 Photometric Quantities Quantity Definition Units Luminous flux That quantity of radiant flux which expresses its capacity to lumen (lm) produce visual sensation Luminous intensity The luminous flux emitted in a very narrow cone containing the candela (cd) given direction divided by the solid angle of the cone (i.e., luminous flux/unit solid angle) Illuminance The luminous flux/unit area at a point on a surface lumen meter 2 Luminance Reflectance For a matte surface Luminance factor For a nonmatte surface for a specific viewing direction and lighting geometry The luminous flux emitted in a given direction divided by the product of the projected area of the source element perpendicular to the direction and the solid angle containing that direction, i.e. luminous flux/unit solid angle/unit area The ratio of the luminous flux reflected from a surface to the luminous flux incident on it: illuminance reflectance luminance = π The ratio of the luminance of a reflecting surface, viewed in a given direction to that of a perfect white uniform diffusing surface identically illuminated: illuminance luminance factor luminance = π (lm m 2 ) candela meter 2 (cd m 2 ) Table 2 Common Photometric Units of Measurement for Illuminance and Luminance and the Factors Necessary to Change Them to SI Units Quantity Unit Dimensions Multiplying Factor to Convert to SI Unit Illuminance (Sl unit = lumen meter 2 ) lux lumen meter meter candle lumen meter phot lumen centimeter 2 10, foot candle lumen foot Luminance (SI unit = candela meter 2 ) nit candela meter stilb candela centimeter 2 10, candela inch 2 1, candela foot apostilb a lumen meter blondel a lumen meter lambert a lumen centimeter 2 3, foot-lambert a lumen foot a These four items are based on an alternative definition of luminance. This definition is that if the surface can be considered as perfectly matte, its luminance in any direction is the product of the illuminance on the surface and its reflectance. Thus, the luminance is described in lumens per unit area. This definition is deprecated in the SI system. Table 3 Situation Typical Illuminance and Luminance Values Illuminance on Horizontal Surface (lm m 2 ) Typical Surface Luminance (cd m 2 ) Clear sky in summer in northern temperate zones Grass 2900 Overcast sky in summer in northern temperate zones Grass 300 Textile inspection 1500 Light gray cloth 140 Office work 500 White paper 120 Heavy engineering 300 Steel 20 Good street lighting 10 Concrete road surface 1.0 Moonlight 0.5 Asphalt road surface 0.01

4 646 EQUIPMENT, WORKPLACE, AND ENVIRONMENTAL DESIGN White 10 Value Scale 9 10B 5PB 5P 10P Hue Scale 5B 8 5RP 10BG 5BG RP 5R G R 5G Chroma Scale 10GY 10YR 5YR 10Y 5GY 2 5Y 1 Black Figure 2 Organization of the Munsell color system. The hue letters are B, blue; PB, purple/blue; P, purple; RP, red/purple; R, red; YR, yellow/red; Y, yellow; GY, green/yellow; G, green; BG, blue/green. need and to provide a more accurate characterization of color, the CIE has developed a system of colorimetry ranging from the complex to the relatively simple (CIE, 1971, 1972, 1978, 1986, 1995, 1998a). The most fundamental characteristic of light is its spectral power distribution reaching the eye. It is this spectral power distribution that largely determines the color seen. Unfortunately, comparisons between spectral power distributions are difficult to comprehend. The CIE has developed two three-dimensional color spaces, both based on mathematical manipulations applied to spectral power distributions (Robertson, 1977; CIE, 1978). These two three-dimensional color spaces, L ab and L uv, are the most comprehensive means of quantifying color, the L ab space being used mainly for object colors and the L uv space being used mainly for self-luminous colors. If two colors have the same coordinates in one of these color spaces, under the same observing conditions they will appear the same. The distance two colors are apart in color space is related to how easily they can be distinguished. An earlier CIE color space, the 1964 Uniform Color Space, is used in the calculation of the CIE General Color Rendering Index, a single-number index which is applied to light sources to indicate how accurately they render colors relative to some standard (CIE, 1995). Specifically, the positions in color space of eight test colors, under a reference light source and under the light source of interest, are calculated. The separation between the two positions of each test color are calculated, and the separations for all the test colors are summed and scaled to give a value of 100 when there is no separation for any of the test colors (i.e., for perfect color rendering). It should be noted that this is a very crude system. Different light sources have different reference light sources, and the summation means that light sources that render the test colors differently can have the same Color Rendering Index. Nonetheless, the Color Rendering Index is widely used as a means of classifying the color-rendering capabilities of light sources. Three two-dimensional color surfaces are still widely used to characterize the color appearance of light sources and to define the acceptable color characteristics of light signals (CIE, 1994). The most commonly used color surface is the CIE 1931 chromaticity diagram (CIE, 1971) (Figure 3). Essentially, it is a slice through color space at a fixed luminance. The curved boundary of the chromaticity diagram consists of the colors produced by single wavelengths. The equal energy point in the center of the diagram corresponds to a colorless surface. The farther the coordinates of a color are from the equal energy point and the closer they are to the boundary, the greater the strength of the color. Figure 3 also shows several areas in which a signal light needs to fall if it is to be perceived as the color specified. The color appearance of light sources is conventionally described by their correlated color temperature. This is the temperature of the full radiator that is closest to the coordinates of the light source on the CIE 1931 chromaticity diagram (Wyszecki and Stiles, 1982). The two other two-dimensional chromaticity diagrams are the CIE 1960 and the CIE 1976 Uniform Chromaticity Scale diagrams. These are linear transformations of the CIE 1931 chromaticity diagram intended to make the surface more perceptually uniform. Whenever chromaticity coordinates are quoted, care should be taken to state the chromaticity diagram being used. A useful summary of these colorimetry systems is given in the 9th edition of the Lighting Handbook of the Illuminating Engineering Society of North America (IESNA, 2000). 2.3 Instrumentation The instrumentation for measuring photometric and colorimetric quantities can be divided into laboratory and field equipment. Laboratory equipment tends to be

5 ILLUMINATION 647 y GREEN YELLOW WHITE RED BLUE 0.1 VIOLET x Figure 3 CIE 1931 chromaticity diagram. The boundary curve is the spectrum locus with the wavelengths (nm) marked. The filled circle is the equal energy point. The enclosed areas indicate the chromaticity coordinates of light signals that will be identified as the specified colors. large and/or sophisticated and hence expensive. Field equipment is small and portable. The luminous flux from a light source, the luminous intensity distribution of a luminaire, and light source color properties are measured conventionally in the laboratory. The two most commonly occurring field instruments are the illuminance meter and the luminance meter. Illuminance meters have three important characteristics: sensitivity, color correction, and cosine correction. Sensitivity refers to the range of illuminances covered, the range desired being dependent on whether the instrument is to be used to measure daylight, interior lighting, or nighttime exterior lighting. Color correction means that the illuminance meter has a spectral sensitivity matching the CIE standard photopic observer. Cosine correction means that the illuminance meter s response to light striking it from directions other than the normal follows a cosine law. The luminance meter is designed to measure the average luminance over a specified area. The luminance meter has an optical system that focuses an image on a detector. Looking through the optical system allows the operator to identify the area being measured and usually displays the luminance of the area. The important characteristics of a luminance meter are its spectral response, its sensitivity, and the quality of its optical system. Again a good luminance meter has a spectral response matching the CIE standard photopic observer. The sensitivity needed depends on the conditions under which it will be used. The quality of its optical system can be measured by its sensitivity to light from outside the measurement area (CIE, 1987). Recently, imaging photometers have become more widely available (Rea and Jeffrey, 1990; Ashdown and Franck, 1995). These instruments are based around a digitized image captured from a video camera. Such instruments are expensive but do provide a means for measuring the luminance of detailed or rapidly changing scenes. Procedures for using illuminance or luminance meters in the field and for light measurements in the laboratory are described and referenced in the guidance published by national bodies (IESNA, 2000; CIBSE, 2002). It should be noted that virtually all commercial instrumentation used to measure illuminance and luminance uses the CIE standard photopic observer as the basis of the instrument s spectral sensitivity, even when the instrument is designed to be used in mesopic and scotopic conditions. 3 PRODUCTION OF ILLUMINATION Illumination is produced naturally, by the sun and artificially, by electric light sources. The development and growth in use of electric light sources over the last

6 648 EQUIPMENT, WORKPLACE, AND ENVIRONMENTAL DESIGN century has fundamentally changed the pattern of life for everyone. 3.1 Daylight, Sunlight, and Skylight Natural light is light received on Earth from the sun, either directly or after reflection from the moon. The prime characteristic of natural light is its variability. Natural light varies in magnitude, spectral content, and distribution with different meteorological conditions, at different times of day and year, at different latitudes. Moonlight is of little interest as a source of illumination but daylight is used, and strongly desired, for the lighting of buildings. Daylight can be divided into two components, sunlight and skylight. Sunlight is light received at Earth s surface, directly from the sun. Sunlight produces strong, sharp-edged shadows. Skylight is light from the sun received at Earth s surface after scattering in the atmosphere. Skylight produces only weak, diffuse shadows. The balance between sunlight and skylight is determined by the nature of the atmosphere and the distance that the light passes through it. The greater the amount of water vapor and the longer the distance, the higher is the proportion of skylight. The illuminances on Earth s surface produced by daylight can cover a large range, from 150,000 lx on a sunny summer s day to 1000 lx on a heavily overcast day in winter. Several models exist for predicting the daylight incident on a plane, at different locations, for different atmospheric conditions (Robbins, 1986). These models can be used to predict the contribution of daylight to the lighting of interiors. The spectral composition of daylight also varies with the nature of the atmosphere and the path length through it. The correlated color temperature of daylight can vary from 4000 K for an overcast day to 40,000 K for a clear blue sky. For calculating the appearance of objects under natural light, the CIE recommends the use of one of three different spectral distributions corresponding to correlated color temperatures of 5503, 6504, and 7504 K (Wyszecki and Stiles, 1982). 3.2 Electric Light Sources The lighting industry makes several thousand different types of electric lamps. Those used for providing illumination can be divided into two classes: incandescent lamps and discharge lamps. Incandescent lamps produce light by heating a filament. Discharge lamps produce light by an electric discharge in a gas. Incandescent lamps operate directly from mains electricity. Discharge lamps all require control gear between the lamp and the electricity supply, because different electrical conditions are required to initiate the discharge and to sustain it. Electric light sources can be characterized on several different dimensions. They are: Luminous efficacy: the ratio of luminous flux produced to power supplied (lumens watt 1 ). If the lamp needs control gear, the watts supplied should include the power demand of the control gear. Correlated color temperature: a measure of the color appearance of the light produced, measured in degrees Kelvin (see Section 2.2.2). CIE general color rendering index: a measure of the ability to render colors accurately (see Section 2.2.2). Lamp life: the number of burning hours until either lamp failure or a stated percentage reduction in light output occurs. Lamp life can vary widely with switching cycle. Run-up time: the time from switch-on to full light output. Restrike time: the time delay between the lamp being switched off before it will reignite. Table 4 summarizes these characteristics for two incandescent lamp types and six discharge lamp types that are widely used and gives the most common applications for each lamp type. The values in Table 4 should be treated as indicative only. Details about the characteristics of any specific lamp should always be obtained from the manufacturer. Many of the lamp types described in the table are also used for internally and externally illuminated signs and signals, but there are other lamp types, operating on different principles, that are used for this purpose (Boyce, 2003). Of these, the one of most interest is the light-emitting diode (LED). The LED has become the lamp type of choice for traffic signals and exit signs. The LED is a semiconductor that emits light when a current is passed through it. The spectral emission of the LED depends on the materials used to form the semiconductor, and typically is narrowband, giving a highly saturated color. LEDs have been developed to produce white light, either by combining red, green, and blue LEDs or by attaching a phosphor to a blue LED. White LEDs currently have a luminous efficacy comparable with an incandescent lamp, but a much longer life and greater durability. They are being used as sources of local illumination (e.g., aircraft reading lights). Considerable time and money is being devoted to enhancing the luminous efficacy and reducing the cost of white LEDs so they are competitive with other lamp types for general illumination. 3.3 Control of Light Distribution Being able to produce light is only part of what is necessary to produce illumination. The other part is to control the distribution of light from the light source. For daylight, this is done by means of window shape, placement, and glass transmittance (Robbins, 1986). For electric light sources, it is done by placing the light source in a luminaire. The luminaire provides electrical and mechanical support for the light source and controls the light distribution. The light distribution is controlled by using reflection, refraction, or diffusion, individually or in combination (Simons and Bean, 2000). One factor in the choice of which method of light control to adopt in a luminaire is the balance desired between the reduction in the luminance of the

7 ILLUMINATION 649 Table 4 Source Properties of Some Widely Used Electric Light Sources Luminous Efficacy (lm/w) Correlated Color Temperature (K) CIE General Color Rendering Index Lamp Life (hr) Run-up Time (min) Restrike Time (min) Applications Incandescent Tungsten Instant Instant Residential, retail Tungstenhalogen Instant Instant Display Discharge Low-pressure , Instant Commercial mercury (fluorescent lamp) Compact , Instant Commercial, retail fluorescent lamp High-pressure mercury (vapor) ,000 24, Older industrial agricultural High-pressure mercury (metal , Industrial, commercial, retail halide) Low-pressure n/a 16,000 18, Security, road sodium High-pressure sodium ,000 24, Industrial, road light source and the precision required in light distribution. Highly specular reflectors can provide precise control of light distribution, but do little to reduce source luminance. Conversely, diffusers make precise control of light distribution impossible but do reduce the luminance of the luminaire. Refractors are an intermediate case. The light distribution provided by a specific luminaire is quantified by the luminous intensity distribution. All reputable luminaire manufacturers provide luminous intensity distributions for their luminaires. With luminaires, you tend to get what you pay for. Luminaires, well constructed from quality materials, cost more. 3.4 Control of Light Output The control of daylight admitted through a window is achieved by mechanical structures, such as light shelves, or by adjustable blinds (Littlefair, 1990). Whenever the sun, or a very bright sky, is likely to be directly visible through a widow, some form of blind will be required. Blinds can take various forms; horizontal, venetian, vertical, and roller being the most common. Blinds can also be manually operated or motorized, either under manual control or under photocell control. Probably the most important feature to consider when selecting a blind is the extent to which it preserves a view of the outside. Roller blinds that can be drawn down to a position where the sun and/or sky is hidden but the lower part of the widow is still open are an attractive option. Roller blinds made of a mesh material can preserve a view through the whole window while reducing the luminance of the view out. Such blinds are an attractive option where the problem is an overbright sky but will be of limited value when a direct view of the sun is the problem. The same applies to low-transmission glass. For electric light sources, control of light output is provided by switching or dimming systems. Switching systems can vary from the conventional manual switch to sophisticated daylight control systems that dim lamps near windows when there is sufficient daylight. Time switches are used to switch off all or parts of a lighting installation at the end of the working day. Occupancy sensors are used to switch off lighting when there is nobody in the space. Such switching systems can reduce electricity waste but will be irritating if they switch lighting off when it is required, and they may shorten lamp life if switching occurs frequently. The factors to be considered when selecting a switching system are whether to rely on a manual or an automatic system, and if it is automatic, how to match the switching to the activities in the space. If your interest is primarily in reducing electricity consumption, a good principle is to use automatic switch off and manual switch on. This principle uses human inertia for the benefit of reducing energy consumption. If you wish to rely on voluntary manual switching of lighting, care should be taken to make the lighting being switched visible from the control panel and to label the switches so that the operator knows which lamps are being switched. Labels asking people to switch off the lighting when it is not needed can be effective. As for dimming systems, these all reduce light output and energy consumption, but a different system is required for each lamp type. The factors to consider when evaluating a dimming system are the range over

8 650 EQUIPMENT, WORKPLACE, AND ENVIRONMENTAL DESIGN NEAR VISION Lens rounded Sclera Iris contracted Cornea Retina Fovea Pupil Iris opened Lens flattened DISTANT VISION Ciliary muscle Blind spot Optic nerve Retina Lateral geniculate body Optic chiasma Optic Nerve Visual area of cortex Figure 4 system. Section through the eye adjusted for near and distant vision, and the binocular nerve pathways of the visual which dimming can be achieved without flicker or the lamp extinguishing, the extent to which the color properties of the lamp change as the light output is reduced, and any effect that dimming has on lamp life and energy consumption. Sophisticated lighting control systems are available for some light sources which allow the user to have a number of preset scenes. These systems use dimming and switching to alter the lighting of a space. They are commonly used in rooms with multiple functions, such as conference rooms. 4 FUNCTIONAL CHARACTERISTICS OF THE HUMAN VISUAL SYSTEM 4.1 Visual System Structure Illumination is important to humans because it alters the stimuli to the visual system and the operating state of the visual system itself. Therefore, an understanding of the capabilities of the visual system and how they vary with illumination is important to an understanding of the effects of illumination. The visual system is composed of the eye and brain working together. Light entering the eye is brought to focus on the retina by the combined optical power of the air/cornea surface and the lens of the eye. The retina is really an extension of the brain, consisting of two different types of photoreceptors and numerous nerve interconnections. At the photoreceptors, the incident photons of light are absorbed and converted to electrical signals. The nerve interconnections take these signals and carry out some basic image processing. The processed image is transmitted up the optic nerve of each eye to the optic chiasma, where nerve fibers from the two eyes are combined and transmitted to the left and right parts of the visual cortex. It is in the visual cortex that the signals from the eye are interpreted in terms of past experience (Figure 4). Many of the capabilities of the visual system can be understood from the organization of the retina. The two types of visual photoreceptors, rods and cones from their anatomical appearance, have different wavelength sensitivities and different absolute sensitivities to light and are distributed differently across the retina. Rods are the more sensitive of the two and effectively provide a night retina. Cones are less sensitive to light and operate during daytime. In fact, there are three types of cones, each with a different spectral sensitivity. These cones are commonly called

9 ILLUMINATION 651 Density (thousands mm 2 ) cones rods nasal retina fovea temporal retina Angular eccentricity (degrees) Figure 5 Density of rod and cone photoreceptors across the retina on a horizontal meridian. (After Osterberg, 1935.) long-, middle-, and short-wavelength cones, from their regions of maximum spectral sensitivity. These three cone types combine to give the perception of color. Figure 5 shows the distribution of rods and cones across the retina. Cones are concentrated in a small central area of the retina called the fovea that lies where the visual axis of the eye meets the retina, although there are cones distributed evenly across the rest of the retina. Rods are absent from the fovea, reaching their maximum concentration about 20 from the fovea. This variation in concentration of rods and cones with deviation for the fovea is amplified by the number of photoreceptors connected to each optic nerve fiber. In the fovea, the ratio of photoreceptors to optic nerve fibers is close to 1 but increases rapidly as the deviation from the fovea increases. The net effect of this structure is to provide different functions for the fovea and the periphery. The fovea is the part of the retina that provides fine discrimination of detail. The rest of the retina is devoted primarily to detecting changes in the visual environment which require the attention of the fovea. 4.2 Wavelength Sensitivity The rod and cone photoreceptors have different absolute spectral sensitivities (Figure 6). The spectral response of the cones lies between 380 and 780 nm, with the peak sensitivity occurring at 555 nm. The spectral response of the rods lies between 380 and 780 nm, with the peak at 507 nm. The peak sensitivity of the rods is much greater than that of the cones. These spectral sensitivities form the basis of the CIE standard observers and hence the photometric quantities discussed in Section 2.1. By adjusting the spectral emission of a light source to lie within the most sensitive part of the spectral response of the visual system, lamp manufacturers are able to vary the luminous efficacy of their light sources (i.e., to change the number of lumens emitted for each watt of power applied). Log relative spectral sensitivity Wavelength (nm) rods foveal cones peripheral cones Figure 6 Log relative spectral sensitivity of rod and cone photoreceptors plotted vs. wavelength. (After Wald, 1945.) 4.3 Adaptation The visual system can operate over a range of about 12 log units of luminance, from a luminance of 10 6 to 10 6 cd/m 2, from starlight to bright sunlight. But it cannot cover this range simultaneously. At any instant in time, the visual system can cover a range of two or three log units of luminance. Luminances above this limited range are seen as glaringly bright, those below as undifferentiated black. The capabilities of the visual system depend on where in the complete range of luminances it is adapted. Three different functional ranges of luminance are conventionally identified: the photopic, mesopic, and scotopic. Table 5 summarizes the visual system capabilities in each of these functional ranges. The visual system adjusts its state of adaptation continuously through three mechanisms: neural, mechanical, and photochemical. These three mechanisms differ in their speed and range of adjustment. The neural mechanism, which is based in the retina, operates in milliseconds and covers a range of two to three log units in luminance. The mechanical mechanism involves the expansion and contraction of the iris. The consequent changes in pupil size take about a second but cover less than one log unit in luminance. The photochemical mechanism covers the entire range of luminance but is slow, the changes taking minutes. The exact time will depend on the starting and finishing luminances for the adaptation. If both are greater than 3cd/m 2, only cones are involved. As the time constant for cones is of the order of 2 to 3 minutes, adaptation takes only a few minutes. When the starting luminance is in the operating range of the cones and the finishing luminance is within the operating range of the rods, a two-stage adaptation process occurs, involving both cones and rods. As rods have a time constant around

10 652 EQUIPMENT, WORKPLACE, AND ENVIRONMENTAL DESIGN Table 5 Name Functional Ranges of Visual System Capabilities Luminance Range (cd m 2 ) Photoreceptor Active Wavelength Range (nm) Capabilities Photopic >3 Cones Color vision, good detail discrimination Scotopic <0.001 Rods No color vision, poor detail discrimination Mesopic >0.001 and <3 Cones and rods Diminished color vision, reduced detail discrimination, and a shift in spectral sensitivity as adaptation luminance moves from photopic to scotopic 7 to 8 minutes, the adaptation time is much longer. Complete adaptation from a high photopic luminance to darkness can take up to an hour. Interior lighting is almost always sufficient for the visual system to be operating in the photopic region. Exterior lighting on roads and in urban areas is usually sufficient to keep the visual system operating in the low photopic or mesopic regions. It is in very rural areas, at sea, or underground, where there is neither exterior lighting nor moonlight that the visual system reaches scotopic adaptation. The speed of adaptation is important where a large and sudden change in the luminance occurs. Examples of situations where this happens are the entrance to road tunnels during daytime (Bourdy et al., 1987) and the onset of emergency lighting during a power failure (Boyce, 1985). These problems are overcome either by installing a gradual reduction in luminance which allows more time for adaptation to occur or by setting a minimum luminance within the neural adaptation range. 4.4 Color Vision When photopically adapted, the visual system can discriminate many thousands of colors. This ability to discriminate colors reduces as the adaptation luminance decreases through the mesopic region and vanishes in the scotopic vision. This is because color vision is mediated by the cone photoreceptors. Different light sources have different spectral emissions and hence render colors differently. To ensure good color discrimination, it is necessary to use a light source that has a high CIE General Color Rendering Index and produces sufficient light to ensure that the visual system is operating in the photopic region. However, it is important to note that light sources with the same CIE Color Rendering Index do not necessarily render all colors in the same way. For example, an incandescent lamp and a fluorescent lamp, both of which can have CIE Color Rendering Index values in the 90s, make blue and green colors appear very different. If you are concerned about color appearance as well as color discrimination, you will have to chose a light source that gives both good color discrimination and the desired color appearance. 4.5 Receptive Field Size and Eccentricity The retina is organized such that increasing numbers of photoreceptors are connected to each optic nerve fiber as the deviation from the fovea increases. This feature of the visual system is important when detection of a stimulus is necessary and it can occur anywhere in the visual field. The visual system will normally operate by first detecting the stimulus off-axis (i.e., in the peripheral visual field) and then turning the eye so that the stimulus is brought onto the fovea for detailed examination. To identify a stimulus offaxis, the stimulus should be clearly different from its background, in luminance or color, and should change in space or time (i.e., it should either move or flicker). A flickering light is commonly used to draw drivers attention to important signs placed beside or above the road. 4.6 Meaningful Stimulus Parameters Any stimulus to the visual system can be described by five parameters: visual size, luminance contrast, chromatic contrast, retinal image quality, and retinal illumination. These parameters are important in determining the extent to which the visual system can detect and identify the stimulus Visual Size The visual size of a stimulus describes how big the stimulus is. The larger a stimulus is, the easier it is to detect. There are several different ways to express the size of a stimulus presented to the visual system, but all of them are angular measures. The visual size of a stimulus for detection is best given by the solid angle the stimulus subtends at the eye. The solid angle is given by the quotient of the areal extent of the object and the square of the distance from which it is viewed. The larger the solid angle, the easier the stimulus is to detect. The visual size for resolution is usually given as the angle the critical dimension of the stimulus subtends at the eye. What the critical dimension is depends on the stimulus. For two points, the critical dimension is the distance between the two points. For two lines it is the separation between the two lines. For a Landolt ring, it is the gap size. The larger the visual size of detail in a stimulus, the easier it is to resolve the detail. For complex stimuli, the measure used to express their dimensions is the spatial frequency distribution. Spatial frequency is the reciprocal of the angular subtense of a critical detail, in cycles per degree. Complex stimuli have many spatial frequencies and hence a spatial frequency distribution. The match

11 ILLUMINATION 653 between the spatial frequency distribution of the stimulus and the contrast sensitivity function of the visual system (see Section 5.2) determines if the stimulus will be seen and what detail will be resolved. Lighting can change the visual size of threedimensional stimuli by casting shadows that extend or diminish the apparent visual size of the stimulus Luminance Contrast The luminance contrast of a stimulus quantifies its luminance relative to its background. The higher the luminance contrast, the easier it is to detect the stimulus. There are two different forms of luminance contrast. For stimuli that are seen against a uniform background, luminance contrast is defined as C = L t L b /L b where C is the luminance contrast, L t the luminance of the detail, and L b the luminance of the background. This formula gives luminance contrasts that range from 0 to 1 for stimuli that have details darker than the background and from 0 to infinity for stimuli that have details brighter than the background. It is widely used for the former (e.g., printed text). For stimuli that have a periodic pattern (e.g., a grating), the luminance contrast or modulation is given by C = (L max L min )/(L max + L min ) where C is the luminance contrast, L max the maximum luminance, and L min the minimum luminance. This formula gives luminance contrast that ranges from 0 to 1. Lighting can change the luminance contrast of a stimulus by producing disability glare in the eye or veiling reflections from the stimulus or by changing the incident spectral radiation when colored stimuli are involved Chromatic Contrast Luminance contrast uses the total amount of light emitted from a stimulus and ignores the wavelengths of the light emitted. It is the wavelengths emitted from the stimulus that largely determine its color. It is possible to have a stimulus with zero luminance contrast that can still be detected because it differs from its background in color (i.e., it has chromatic contrast). There is no widely accepted measure of chromatic contrast, although various suggestions have been made (Tansley and Boynton, 1978). Fortunately, chromatic contrast becomes important for detection only when luminance contrast has reached a low level. Lighting can alter chromatic contrast by using light sources with different spectral emission characteristics Retinal Image Quality As with all image-processing systems, the visual system works best when it is presented with a clear, sharp image. The sharpness of the stimulus can be quantified by the spatial frequency distribution of the stimulus: A sharp image will have high spatial frequency components present; a blurred image will not. The sharpness of the retinal image is determined by the stimulus itself, the extent to which medium through which it is transmitted scatters light, and the ability of the visual system to focus the image on the retina. Lighting can do little to alter any of these factors, although it has been shown that light sources that are rich in the short wavelengths produce smaller pupil sizes, and these tend to improve visual acuity for briefly presented low-contrast targets. The explanation suggested is that the smaller pupil sizes produce greater depth of field and hence better retinal image quality (Berman et al., 1993) Retinal Illumination The retinal illumination determines the state of adaptation of the visual system and therefore alters its capabilities. The retinal illumination is determined by the luminance in the visual field, modified by pupil size. Retinal illumination is measured in trolands, a quantity formed from the product of the luminance of the visual field and the pupil size (Wyszecki and Stiles, 1982). Illuminances and surface reflectances determine the luminances of the visual field. Luminances and light spectrum determine pupil size. 5 EFFECTS ON THRESHOLD VISUAL PERFORMANCE Qualitatively, threshold visual performance is the performance of a visual task close to the limits of what is possible. Quantitatively, it is the performance of a task at a level such that it can be carried out correctly on 50% of the occasions it is undertaken. Threshold visual performance is affected by many different variables. For example, visual acuity is affected by the form of the target used, the luminance contrast of the target, the duration for which it is presented, where in the visual field it appears, and the luminance of the surround relative to the luminance of the immediate background. In this discussion of threshold visual performance, attention will be limited to the effects of variables that are controlled by the lighting system (i.e., the adaptation luminance and the spectral content of the light). Information on the influence of other variables can be obtained from Boff and Lincoln (1988). In the data presented it will be assumed that the observer is fully adapted to the prevailing luminance, that the image of the target is on the fovea, that the target is presented for an unlimited time, and that the observer is correctly refracted. Again, the influence of departures from these assumptions can be estimated from the data given by Boff and Lincoln (1988). 5.1 Visual Acuity Visual acuity is the limit in the ability to resolve detail. Visual acuity has frequently been measured using gratings or Landolt C s. Visual acuity can be quantified as the angle subtended at the eye by the size of detail that can be detected correctly on 50% of the occasions it is presented. No matter what target is used, visual acuity improves (i.e., the size of detail that can be resolved decreases) as adaptation luminance

12 654 EQUIPMENT, WORKPLACE, AND ENVIRONMENTAL DESIGN Visual acuity (min arc) log luminance (cd/m 2 ) Figure 7 Effect of adaptation luminance on the gap size of a Landolt C target which can just be resolved. (After Shlaer, 1937.) increases. Figure 7 shows that as adaptation luminance increases from scotopic to photopic conditions, the visual acuity increases, asymptotically approaching a maximum at high luminances. The adaptation luminance produced by a lighting installation will depend on the illuminances produced on different surfaces and the reflectance of those surfaces. Table 3 gives some luminances typically found in interior and exterior lighting installations. Given a value for the adaptation luminance, Figure 7 can be used to determine if detail of a given size can be resolved. A useful rule of thumb is that the detail needs to be four times bigger than the visual acuity limit if it is to be resolved sufficiently quickly to avoid affecting visual performance (Bailey et al., 1993). As for the light spectrum, provided that the lamp produces white light rather than an emission in a narrow spectral region, the effect on visual acuity is very small, certainly much less than the effect of adaptation luminance (Shlaer et al., 1941). 5.2 Contrast Sensitivity Function Contrast sensitivity is the reciprocal of the luminance contrast that can be detected on 50% of the occasions it is presented. Contrast sensitivity is usually measured using a sinusoidal grating target. The contrast sensitivity function is contrast sensitivity plotted against the spatial frequency of the sinusoidal target. Figure 8 shows the effect of adaptation luminance on the contrast sensitivity function. It shows that as the adaptation luminance increases from scotopic to photopic conditions, the contrast sensitivity increases for all spatial frequencies; the spatial frequency at which the peak contrast sensitivity occurs increases and the highest spatial frequency that can be detected also increases. Figure 8 can be used to determine if a given target will be visible by breaking the target into its spatial frequency components and determining if any of the components are within the limit set by the contrast sensitivity function (Sekular and Blake, 1994). The target will only be visible if at least one of its components falls within this limit, although it should be noted that the appearance of the target will be different depending on which component or components are visible. As a rule of thumb, for a target to be easily seen, it is necessary for the luminance contrast to be at least twice the contrast threshold. As for the light spectrum, there is no evidence that the contrast sensitivity function is influenced by different white light spectra, provided that the luminances are the same. 5.3 Temporal Sensitivity Function The temporal sensitivity function shows percentage modulation amplitude plotted against the frequency of the modulation. Figure 9 shows the effect of adaptation luminance on the temporal sensitivity function. It shows that as the adaptation luminance increases Contrast Sensitivity Luminance (cd/m 2 ) and Spatial Frequency (cycles/degree) Figure 8 Effect of adaptation luminance on the contrast sensitivity function. Contrast sensitivity is plotted vs. spatial frequency (cycles/degree) for various adaptation luminances. (After Boff and Lincoln, 1988.)

13 ILLUMINATION 655 % Modulation amplitude Frequency (Hz) Figure 9 Effect of adaptation luminance on the temporal sensitivity function. Percentage modulation amplitude is plotted vs. frequency (in hertz) for various levels of retinal illumination. The retinal illuminations are: filled square, 0.06 troland; open square, 0.65 troland; open, inverted triangle, 7.1 trolands; open, upright triangle, 77 trolands; open circle, 850 trolands; filled circle, 9300 trolands (After Kelly, 1961.) from mesopic to photopic conditions, the temporal sensitivity increases for all frequencies; the frequency at which the peak temporal sensitivity occurs increases, and the highest frequency that can be detected also increases. Figure 9 can be used to determine if a given temporal variation will be visible by breaking the waveform representing the light fluctuation into its frequency components and determining if any of the components are within the limit set by the temporal sensitivity function. The fluctuation will be visible only if at least one of its frequency components falls within this limit. Temporal fluctuation in luminous flux (i.e., flicker) is undesirable in lighting installations. To eliminate flicker, it is necessary to increase the frequency and/or decrease the percentage modulation sufficiently to take their combination outside the limits set by the temporal sensitivity function. In practice, this is easily done. Incandescent lamps have sufficient thermal inertia to ensure that even though the frequency of the fluctuation is only twice the supply frequency (120 Hz for a 60-Hz electrical supply), the percentage modulation is small, so there is little chance of seeing flicker from such a lamp. Discharge lamps, such as the fluorescent lamp, do not have thermal inertia, so their percentage modulation can be high. To ensure that fluorescent lamps do not produce visible flicker, it is best to use an electronic ballast to control the lamp. Electronic ballasts typically operate at frequencies in the tens of kilohertz, with small percentage modulations, and consequently, are very unlikely to produce visible flicker. 5.4 Color Discrimination The ability to discriminate between two colors of the same luminance depends on the difference in spectral power distribution of the light received at the eye. Figure 10 shows the MacAdam ellipses, the area around a number of chromaticities, each magnified 10 times, within which no discrimination of color can be made, even under side-by-side comparison conditions (Wyszecki and Stiles, 1982). The effect of illuminance on the ability to discriminate between colors is limited in the photopic region, an illuminance of 300 lx being sufficient for good color judgment work (Cornu and Harlay, 1969). As the visual system enters the mesopic region, the ability to discriminate colors deteriorates and ultimately fails as the scotopic region is reached. The effect of light spectrum is much more important. The position of a color on the CIE 1931 Chromaticity Diagram is determined by the spectrum of the light, and if it is reflected from or transmitted through a surface, the spectral reflectance or transmittance of that surface. Therefore, by changing the light spectrum emitted by the lamp, it is possible to make colors easily discriminable or difficult to discriminate. The careful choice of light source is important wherever good color discrimination is important. 5.5 Interactions The fact that there are many other variables besides adaptation luminance and light spectrum that influence threshold visual performance has been mentioned earlier. It is now necessary to introduce another complication: interaction between the various components of visual system performance. As an example, consider the effect of luminance contrast on visual acuity. Visual acuity is conventionally measured using targets with a high luminance contrast. However, as the luminance contrast of the target is decreased, visual acuity also worsens. Similarly, the temporal sensitivity function as presented applies to a uniform luminance field. If the field has a pattern and hence a distribution of spatial frequencies, the temporal sensitivity function may be changed (Koenderink and Van Doorn, 1979). Put crudely, what this means is that as visual performance gets closer to threshold, almost everything about the stimulus presented to the visual system becomes important. Further details on some of the interactions that occur are given in Boff and Lincoln (1988). 5.6 Approaches to Improving Threshold Visual Performance Working close to threshold is not easy. In fact, it can be argued that the main function of anyone designing lighting is to provide conditions that avoid the need to use the visual system close to threshold. However, if this is the situation, the following steps can be taken to improve threshold visual performance. Not all of the following steps will be possible in every situation, and not all are appropriate for every problem. The discussion above should indicate which approach is likely to be most effective. Changing the Task Increase the size of the detail in the task.