Speed and sensitivity
SENSITIVITY OF IMAGING SYSTEMS
Spectral sensitivity describes a system’s ability to detect light with respect to wavelength. As spectral sensitivity increases for a given wavelength, the probability of recording the light present at that wavelength increases (Figure 20.1). The vast majority of imaging systems are not equally sensitive to all wavelengths of light and the measure is important, as it will define the scope of use and performance of an imaging system, e.g. the ability to distinguish colour or its viability as a night vision system. Absolute radiometric spectral sensitivity is a primary objective imaging measure, from which many of the others described throughout this book may be derived.
Figure 20.1 An example of a relative spectral sensitivity curve.
The spectral sensitivity of a system may be specified in many forms and it is useful to verify that equivalent methods are being used when comparing measurements. In all cases some measure of sensitivity is plotted against wavelength, as in Figure 20.1. Spectral curves may be presented as relative or absolute measures of sensitivity. In the case of relative sensitivity, the spectral curve is normalized to a unit area, or the peak spectral response to a value of 1. This is by far the most accessible type of curve and can provide a great deal of information. Measurement requires little more than a light source for which the relative spectral output is known and some method of division of the source into appropriate bands of wavelengths with which to expose the sensor, such as a monochromator or spectral wedge. Relative spectral curves are unitless because of the normalization and therefore care must be taken when comparing them. The relative curve represents the response of the system to a ‘unit’ of light; it does not specify what that unit is. Therefore, it is possible for systems to have identical relative spectral sensitivities, though overall one may be half as sensitive to light as the other. This would result in increased noise and a decreased response in the less sensitive system for equivalent exposures.
One may overcome the above problem by producing absolute spectral sensitivity curves. These relate input energy to output units (sometimes in a quite abstract form), from which the absolute sensitivity of the system at each wavelength is determined. Many of the measures detailed in Chapter 24, such as noise equivalent exposure (NEE), saturated equivalent exposure (SEE), responsivity or detective quantum efficiency (DQE) may be used to produce such curves if plotted with respect to wavelength. When comparing the imaging systems in the example above using such measures, it would be readily apparent if one was half as sensitive as the other.
Spectral sensitivity curves may be produced using radiometric or photometric units. Radiometric units, as seen in Chapter 2, describe electromagnetic radiation in terms of its absolute energy using units such as joules or watts. These units are useful for describing imaging systems that do not necessarily use visible light, e.g. an infrared (IR) security camera or an X-ray machine. The absolute spectral sensitivity curve may be specified in a large variety of units depending upon the measure chosen to express the sensitivity and how appropriate it is for the system under consideration. For example, a radiometric calculation of noise equivalent exposure for a digital sensor may use units of VW−1 nm−2. This may be translated as the number of volts produced by the sensor per watt of energy falling on a square micrometre of the detector surface per nanometre wavelength, when plotted as a curve.
Photometric units are much more applicable to systems utilizing visible light. Light is normalized with respect to the luminous efficiency of the eye (see V(λ) curve in Chapters 2 and 5), and is described by units such as lux, lumens and candelas per unit area. Care must be taken, however, if the system or source under evaluation has any significant component outside of the 380–750 nm range used in normalization as the real response will deviate significantly from the normalized calculation. For example, the sensitivity of video cameras is often specified in terms of the lowest number of lux with which it is possible to make a recording. If the IR response of the camera is significant, the voltage measured per unit of light falling on the sensor will be significantly higher than that calculated using the photometric units.
As for many other aspects of imaging, multi-channel or colour systems may be evaluated by measuring each channel individually.
Spectral sensitivity of digital cameras
Many of the mechanisms which define the spectral sensitivity of charge-coupled device (CCD) and complementary metal oxide semiconductor (CMOS) cameras are detailed in Chapter 9. To summarize, the largest contribution to spectral sensitivity in digital cameras is that of the silicon substrate itself, modified by the effects of the overlying metal and polysilicon layers, which attenuate shorter wavelengths. The degree of attenuation of wavelengths below 400 nm is predominantly influenced by the choice of glass used for micro lenses. The depth of the depletion region can affect sensitivity, as penetration of photons is dependent on wavelength, with longer wavelengths travelling deeper into the substrate. The sensitivity of silicon extends to a wavelength of 1.1 mm and therefore an IR reflecting filter is regularly used in this region. The angle with which light strikes this thin-film reflecting filter changes the cut-off frequency and therefore it can create spectral sensitivity variations across the field of view of a sensor (see Chapter 9). Also, as detailed in Chapter 9, in a single-sensor digital camera, colour discrimination is usually facilitated by the inclusion of a colour filter array (CFA) in front of the sensor, splitting the remaining signal into three broad bands of the spectrum (Figure 20.2; also see Chapter 23).
Figure 20.2 The spectral sensitivity of a typical digital sensor after colour filters have been added.
Measurement of the spectral sensitivity of digital cameras
ISO 17321 provides a standard methodology for the evaluation of the relative spectral sensitivity of digital stills cameras. A monochromator is used to evenly illuminate a surface which fills the field of view of the camera. The device filters light to yield a narrow wavelength band in a desired range (see Chapter 5). It is often controlled by computer to easily step through a large number of wavelengths automatically. Fall-off of illumination is limited so that the radiance at the edge of the field of view is at least 70% of that at the centre. The standard specifies that the surface should be an integrating sphere with a diffuser placed over the exit port.
Images of the surface are created between 360 and 830 nm in increments of less than 10 nm, though not smaller than the band-pass of the monochromator. Measurements to 1100 nm are encouraged where possible. The band-pass of the monochromator is recommended as being narrower than 5 nm. The procedure for setting up the camera is as for ISO 14524, which describes opto-electronic conversion function (OECF) evaluation (see Chapter 21). Fixed exposures are used, such that the maximum response falls between 50% and 90% for the full range of the camera. White balance and other automated functions of the camera are fixed. A radiance meter is used to capture the relative illumination of the target as a function of wavelength.
The camera OECF measured for each colour channel is used to linearize with respect to input luminance each colour plane of each image captured, after which the average of 64 × 64 pixel blocks are calculated at the centre of each image. This yields the linearized response of the camera for each wavelength. The relative spectral response may then be calculated for each by dividing the result by the surface radiance measured. The results are then normalized so that the area under the green channel sensitivity curve is 1. The camera OECF may then further be used to yield absolute spectral sensitivities if desired.
The ISO standard also describes a method for determining colorimetric response utilizing a reflective test target with tabulated patches.
Determination of the colour sensitivity of an unknown photographic emulsion
The spectral sensitivity of photographic materials (orthochromatic, panchromatic, IR, etc.) has been discussed extensively in Chapter 13. The colour sensitivity of an unknown emulsion can be most readily investigated in the studio by imaging a colour chart consisting of coloured patches with a reference scale of greys. In one such chart it has been arranged that the different steps of the neutral half have the same luminosities as the corresponding parts of the coloured half when viewed in daylight. If an image of the chart is made, the colour sensitivity of the emulsion being tested, relative to that of the human eye, is readily determined by comparing the tonal values of the image of the coloured half with those of the image of the neutral half. The value of the test is increased if a second exposure is made on a sensor of known colour sensitivity, to serve as a basis for comparison. Unexpectedly dark monochrome records of colours will indicate a low sensitivity in the spectral region concerned.
A more precise, yet still quite practical, method of assessing the colour sensitivity of a sensor is to determine the exposure factors of a selection of filters. A selection of three filters, tricolour blue, green and red, may be used. An unexpectedly large filter factor will reveal a spectral region, that passed by the filter concerned, to which the sensor is relatively insensitive. The properties of the filters used may be adjusted to approximate estimations of visual spectral sensitivity. In the laboratory, the spectral sensitivity of an emulsion may be determined as a wedge spectrogram.
Wedge spectrograms
The spectral response of a photographic material is most completely illustrated graphically by means of a curve known as a wedge spectrogram, and manufacturers usually supply such curves for various materials. A wedge spectrogram indicates relative sensitivity and is obtained by exposing the material through a photographic wedge in an instrument known as a wedge spectrograph. The spectrograph produces an image in the form of a spectrum on the material, the wedge being placed between the light source and the emulsion. A typical optical arrangement is shown in Figure 20.3.
Figure 20.3 The optical arrangement for a wedge spectrograph.
Examples of the results obtained in a wedge spectrograph are shown in Figure 13.5 in Chapter 13, which shows wedge spectrograms of typical materials of each of the principal classes of colour sensitivity. The outline of a wedge spectrogram forms a curve showing the relative log sensitivity of the material at any wavelength. Sensitivity is indicated on a logarithmic scale, the magnitude of which depends on the gradient of the wedge employed. All the spectrograms illustrated here were made using a continuous wedge, but step wedges are sometimes used. In Figure 13.5 the short-wave cut-off at the left of each curve is characteristic not of the material, but of the apparatus in which the spectrograms were produced, and arises because of ultraviolet absorption by glass. The curves of the spectrally sensitized materials are also seen to consist of a number of peaks; these correspond to the summed absorption bands of the dyes employed.
The shape of each curve depends not only on the sensitivity of the material, but also on the quality of the light employed. All the spectrograms shown in Figure 13.5 were made with a tungsten light source with a colour temperature of approximately 2856 K. Wedge spectrograms of the same materials, made with daylight, would show higher peaks in the blue region and lower peaks in the red.
The wedge spectrogram of the IR material in Figure 13.5 of Chapter 13 shows a gap in the green region of the spectrum. This permits the handling of the material by a green safe light. IR-sensitive materials do not necessarily have this green gap and furthermore may receive IR radiation transmitted by materials which appear opaque to light. Some black plastics and fabrics, such as camera bellows, may transmit IR radiation, as may the sheaths in dark slides used for the exposure of sheet film in technical cameras. Unexpected fogging of extended sensitivity emulsions or IR materials may occur and constructional materials may be implicated if this is found.
The colour material whose spectral sensitivity is shown in Figure 13.6 of Chapter 13 was designed for daylight exposure and was therefore exposed using a tungsten source of 2856 K and a suitable filter to convert the illumination to the quality of daylight. The omission of the conversion filter would have resulted in a relatively higher red sensitivity peak, at about 640 nm, and a lower blue peak, at about 450 nm.
The necessity to balance the exposing illuminant and the film sensitivity by the use of correction filters can be inconvenient, although generally accepted by professional photographers. The ill-effects of omitting correction filters can be substantially reduced by using a negative film specially designed for the purpose, termed type G or, sometimes, universal film. The sensitivity bands of the type G film are broad and less separated than those of the conventional film. The sensitivity characteristics of type G materials give an acceptable colour reproduction over a wide range of colour temperatures, though the optimum is still obtained with daylight exposure. This tolerance to lighting conditions is gained at the expense of some loss in colour saturation and other mechanisms may have to be used by the manufacturer to restore this to an appropriate level. Alternatively, very fast blue-sensitive emulsions of extra-large latitude may be used to provide a suitable image over a wide range of illuminant quality. Such a film will yield negatives of unusually high blue density when used in daylight and these may prove awkward to print using automatic printers.
Uses of wedge spectrograms
Although they are not suitable for accurate measurements, wedge spectrograms do provide a ready way of presenting information. They are commonly used:
1. To show the way in which the response of an emulsion is distributed through the spectrum.
2. To compare different emulsions. In this case, the same light source must be used for the two exposures. It is normal practice to employ as a source a filtered tungsten lamp giving light equivalent in quality to daylight (approx. 5500 K) or an unfiltered tungsten lamp operating at 2856 K, whichever is the more appropriate.
3. To compare the quality of light emitted by different sources. For this type of test, all exposures must be made on the same emulsion.
SPEED OF IMAGING SYSTEMS
Whilst the spectral sensitivity curves discussed above are useful in a large variety of applications, they do little to aid the photographer to correctly expose images quickly and effectively under varying illumination levels and conditions. In practice an alternative method is needed to determine exposure readily ‘in the field’.
The speed of an imaging system designates the relative amount of light needed to produce an image with a defined ‘quality’ and is expressed as a number which may be related to exposure. High speed numbers indicate a greater sensitivity to light and therefore lower total exposures are needed, i.e. shorter exposure times or smaller apertures. Conversely, lower speed numbers indicate that a larger total exposure is required, leading to longer exposure times and larger apertures. The ‘quality’ of the image may be any one of a number of parameters, as will be seen, such as noise levels, contrast or density (intensity) level.
Methods of expressing speed
The problem of allotting a speed number superficially appears simple. It would seem, for instance, that if we wish to compare the speeds of two systems, all we have to do is to make exposures on each to yield comparable images, and the ratio of the exposures will give us the ratio of the speeds. It is true that we can usefully compare speeds in this way, but the comparison may not be typical of the relation between the two systems under other conditions. Furthermore, it will tell us nothing about the absolute speed of the imaging systems; knowing that one is twice as fast as the other does not give an effective exposure. It is in trying to obtain absolute speed numbers for general application that difficulties arise.
One of the first problems in allotting speed numbers is connected with the variation in the speed of imaging systems with the conditions of use. This problem is generally overcome by agreeing upon standard conditions of exposure and handling for the speed determination. A further problem is concerned with the criterion of exposure to be used as the basis for the measurement of speed, i.e. the particular concept of speed which is to be adopted.
This problem arises from the fact that an image consists not of a single tone but of a range of tones. Consequently, it is not immediately apparent which tonal or pixel value or other attribute should be adopted. The characteristic curve – or the system’s transfer function (see Chapters 8 and 21) – is useful here. Several different points related to this curve have been suggested as speed criteria. Most of these points were derived from the development of photographic imaging and from a historical perspective it is useful to discuss these first. Digital imaging adds unique issues to the definition of speed due to the amount of subsequent image processing that can occur in the system and further standards have been developed to account for these.
Many of the photographic speed points are related to the toe of the characteristic curve, i.e. to the shadow areas of the negative. These criteria can be divided into five main types, as follows: threshold, fixed density, inertia, minimum useful gradient and fractional gradient.
Figure 20.4 The threshold criterion of speed.
Threshold systems
The threshold is the point on the characteristic curve corresponding to a just-perceptible density above fog, i.e. the point where the toe begins. Under the heading ‘threshold systems’ are those systems in which speed is based on the exposure required to give such a density (Figure 20.4). The disadvantages of the threshold as a criterion of speed are that it is difficult to locate exactly and that it is not closely related to the part of the characteristic curve used in practice.
Fixed density
The exposure required to produce a given density above fog may be used as an effective speed point. For general-purpose films, a (diffuse visual) density of 0.1 or 0.2 above fog is frequently selected (Figure 20.5), which corresponds approximately to the density of the deepest shadow of an average negative. With high-contrast materials in which a dense background is required, a density of 1–2 is a more useful basis for speed determination, while for materials used in astronomy, a density of 0.6 has been suggested. It will be apparent that the exposure corresponding to a specified density can be more precisely located than the threshold. A fixed density criterion (D = 0.1 + fog) was adopted in the first National Standard speed system, the DIN system, in 1934, and is now employed in all current systems.
Inertia
This was the basis selected by Hurter and Driffield for their pioneering work on quantifying the photographic process. The inertia point is where an extension of the linear portion of the characteristic curve crosses a line representing the base plus fog level (Figure 20.6). Under the processing conditions which were prevalent in Hurter and Driffield’s time, inertia was independent of development and so offers a fixed point of reference. Further, the inertia point is related to the linear portion of the characteristic curve, i.e. the part of the curve in which objectively correct reproduction in the negative is obtained. With short-toe materials, as used by Hurter and Driffield, this would be an advantage. With modern materials, which have a long toe often with no linear region, the linear portion of the curve has little relevance.
Figure 20.5 Fixed density criterion.
Figure 20.6 Inertia criterion.
Minimum useful gradient
Threshold speed systems work at the very bottom of the toe of the characteristic curve, while systems based on inertia ignore the toe completely. Neither system approximates very closely to actual practice, where a part (but only a part) of the toe is used. It was at one time suggested that a criterion more closely related to practice could be obtained from that point on the toe of the characteristic curve at which a certain minimum gradient is reached. A value of 0.2 for tan a in Figure 20.7 was proposed. The minimum useful gradient criterion was based on the idea that loss of tone separation in the shadows (shadow detail) is the first sign of underexposure, and that this in turn is due to unacceptably low contrast in the portion of the characteristic curve occupied by the shadows. The minimum useful gradient criterion did not come into general use but is of interest because it led to the more fundamental fractional gradient criterion.
Figure 20.7 Minimum useful gradient criterion.
Fractional gradient
The main argument against the minimum useful gradient criterion is that the minimum value of contrast acceptable in the shadows is not a constant but depends upon the contrast grade of the paper on which the negative is to be printed. If the overall contrast of the negative is such that it needs a hard paper, the contrast of the negative in the shadows can be lower than with a negative requiring a soft paper. A hard paper is generally one that produces high image contrast, whereas a soft paper gives low contrast. In other words, the minimum contrast acceptable in the toe depends upon the contrast of the negative as a whole. Realization of this fact led to the conception of the fractional gradient criterion. The point chosen for this criterion is the point A in Figure 20.8, where the slope of the tangent to the curve at A equals a given fraction of the slope of AB, the line joining the points marking the ends of the portion of the curve employed. This is usually expressed by the equation:
where GMin = tan a, 
= tan b (provided the density and log exposure axes are equally scaled) and K is a constant determined empirically.
Figure 20.8 Fractional gradient criterion.
Practical tests by L.A. Jones showed that a value for K of 0.3 gave results corresponding very well with the minimum exposure required to give a negative from which an ‘excellent’ (as opposed to merely ‘acceptable’) print could be made. In Jones’s work, the fractional gradient point A was located by the equation:
where (1:5) means the average gradient over a log exposure range of 1.5, a value which has been shown to be fairly typical for exterior scenes in daylight. When located in this way, point A is sometimes referred to as the ‘Jones point’. This criterion was employed first by Eastman Kodak in 1939, and was later adopted by the then American Standards Association (in 1943) and the British Standards Institution (in 1947) as the basis for national standards.
Speed systems and standards
Table 20.1 outlines standards which were adopted by the various national standards organizations. Speed in all three American, British and German systems is currently determined with reference to the exposure required to produce a density of 0.1 above fog density, this criterion being simpler to use in practice than the fractional gradient criterion. American and British standards were further modified in 1972 and 1973 respectively. Both use the same speed criterion mentioned below but the developer solution and the illuminant specified are slightly different from those specified in the earlier standards. The common method adopted for determining speed in the three standards is illustrated in Figure 20.9. This procedure is used in the current International Organization for Standardization (ISO) standard.
The International Organization for Standardization (ISO) is an international federation of national standard (member) bodies. Its work is undertaken by various technical committees on which interested member bodies have representatives. Draft international standards are adopted when at least 75% of the member bodies have cast supporting votes. It is important to note that where ISO speed values are to be measured, reference must be made to all the specifications, and definitions which can only be found by direct reference to the standard.
Table 20.1 The principal methods of expressing film speed of negative monochrome materials adopted in early standards
Figure 20.9 shows the basic principles now used for the determination of ISO speed of emulsions. The characteristic curve of a photographic material is plotted for the specified developing conditions. Two points are shown on the curve at M and N. Point N lies 1.3 log exposure units from point M in the direction of increasing exposure. The development time of the negative material is so chosen that point N has a density 0.80 greater than the density at point M. When this condition is satisfied, the exposure corresponding to point M represents the criterion from which speed is calculated. It is for the degree of development obtained that the correlation between the fixed density criterion and the fraction gradient criterion that was referred to above holds good. In the current standards speeds are determined from the following formulae:
Figure 20.9 Principles of the method adopted for determining speed in the current ISO standard.
Arithmetic speed
Logarithmic speed
where Hm is the exposure in lux-seconds corresponding to the point M in Figure 20.9. Table 13.3 in Chapter 13 shows equivalent arithmetic and logarithmic speed numbers. The range of exposure values for rounding to the particular speed value are provided in tables of the ISO Standard. In Table 20.1 and elsewhere in the text, reference has been made to arithmetic and logarithmic speed ratings. With arithmetic scales a doubling of film speed is represented by a doubling of speed number. With logarithmic speed scales, distinguished by a degree (°) sign, the progression is based on the common logarithm (base 10 scale). The common logarithm of 2 is almost exactly 0.3, and logarithmic film speeds are scaled so that a doubling of film speed is represented by an increase of 3 in the speed index.
Speed ratings are not published for high-contrast materials, such as those used in the graphic arts or for micro-copying, or for special materials such as astronomical plates, holographic materials and recording films. As we have already seen, any system for expressing the speed of a photographic material must take into account exposure and development conditions, and must be related to some particular criterion of correct exposure. The systems in general use for ordinary photographic materials cannot be applied to high-contrast or special materials since the conditions of exposure and development are quite different. For these applications, speed points are often measured at fixed densities greater than 0.1.
ISO speed ratings for emulsions should not be confused with manufacturers’ exposure indices, which are suggested values for use with exposure meters calibrated for ISO speeds and have been arrived at by manufacturers’ testing procedures for determining camera settings. A speed rating preceded by the letters ‘ISO’ implies that this rating was obtained by exposing, processing and evaluating exactly in accordance with the ISO standard.
ISO SPEED RATINGS FOR COLOUR SILVER-BASED MATERIALS
Colour negative film
The principles of speed determination for colour negative (print) films follow those described previously for monochrome materials. However, speed determination is complicated by the fact that colour materials contain layers sensitive to blue, green and red light, and the standards involve averaging the speeds of the three sensitive layers. The ISO standard (ISO 5800:1987) uses fixed-density criteria for locating the speed points of the three layers, exposed in a defined manner and processed according to the manufacturer’s recommendations after storage at 23 ± 2°C for between 5 and 10 days, for general-purpose films. Blue, green and red densities are measured according to defined ISO measurement standards and the characteristic curves shown in Figure 20.10 are plotted.
The points B, G and R are located at densities of 0.15 above the minimum densities for the blue-, green- and red-sensitive layers. The log exposure values corresponding to these points are log HB, log HG and log HR respectively. The mean exposure, Hm, is then calculated from two of these values (HG and the slowest layer, usually the red-sensitive layer, HR) according to the following formula:
Figure 20.10 Principles of the method for determining speed of colour negative films (BS 1380, Part 3:1980, and used currently in ISO 5800:1987).
or
The logarithmic speed is calculated as follows:
where all exposure values are in lux seconds. The speed values are then determined from a table of log10 Hm provided in the standard. The arithmetic speed may be calculated from the value of Hm by the formula:
Colour reversal film
The standard for colour reversal films involves measurement of diffuse visual density and the plotting of a single characteristic curve, shown in Figure 20.11. Films are exposed in a defined manner and processed according to the manufacturer’s recommendations. For the determination of speed, two points (T and S) are located on the characteristic curve of Figure 20.11. Point T is 0.20 above the minimum density. Point S is that point on the curve at which the density is 2.0 above the minimum density. The log exposure values (log10 HS and log10 HT) corresponding to points S and T are read from the curve and their mean value (Hm) is calculated:
Figure 20.11 Principles of the method for determining the speed of colour reversal (slide) films used in ISO standard ISO 2240:1994(E).
or:
from which the speeds may be calculated by analogous procedures to those used for the previously described ISO standards:
Arithmetic speed
Logarithmic speed
All exposure values are in lux seconds.
SPEED OF DIGITAL SYSTEMS
Evaluation of the speed of digital systems poses particular problems as the degree to which they may be manipulated subsequent to exposure by mathematical processing such as data compression or analogue-to-digital conversion. Their ability to alter speed settings electronically by increasing electrical gain may also bias the result. However, there is also a need for speed ratings for digital systems due to their proliferation in recent years and for the same reasons as for photographic materials. A number of ISO standards have been adopted for the evaluation of various parameters of digital systems.
ISO 12232 was first adopted to rate the speed of digital stills cameras in 1998 and has subsequently been updated in 2006 with the addition of recommended exposure index (REI) and standard output sensitivity (SOS). Similarly to standards for photographic media, those for digital systems specify conditions of illumination, temperature, relative humidity and exposure time. In addition, those variables specific to digital systems are specified. These include:
• Adjustment of white balance for the illumination used, as equal RGB signals according to ISO 14524, Chapter 21.
• Exposure time (photosite integration time) not longer than 1/30 s.
• IR-absorbing (blocking) filter should be used if required (ISO 14524, Chapter 21).
• If the device uses any form of lossy compression this must be disabled.
• If lossy compression systems cannot be disabled then noise-based values for ISO speed determination should not be reported.
• Factory defaults for other settings such as sharpness and contrast filtering, etc.
The standard (ISO 12232) provides four methods for the determination of speed and has as one of its aims a correlation between the ISO speed rating of an electronic camera and that for conventional photographic media. This means that a setting of ISO 200 on an electronic camera requires the same exposure time and aperture to that of a photographic film with an ISO speed of 200.
Speed based on saturation (Ssat)
This is based on the user setting the camera exposure such that the highlights of the image are just below the maximum value (saturation) for the signal value, i.e. before they become clipped. For focal plane measurement of exposure this is defined as:
where Hsat is the minimum exposure level (lxs) that produces the maximum output signal with no clipping.
If exposure cannot be measured in the focal plane, Hsat is calculated as follows:
where Aeff is the effective f-number of the lens, La is the arithmetic mean luminance of the scene expressed in candelas per square metre and t is the photosite integration time (s).
Speed based on noise (Snoise)
This is based on the minimum exposure value before the onset of noise becomes objectionable. This depends on subjective assessments of images and their correlations with noise to decide on the acceptability of noise. Two noise-based speeds are used (Snoise10 and Snoise40), where 10 corresponds to the ‘first acceptable’ image and 40 to the ‘first excellent’ image, determined from psychophysical experiments. For focal plane measurement of exposure this is defined as:
In the above equation, the exposure providing the camera signal-to-noise ratio which must satisfy a number of criteria specified in the standard, x, has the value of 10 or 40. The signal-to-noise criterion may be fulfilled by satisfying the condition:
where x takes the value 40 or 10 as before, D is linearized signal level and s(D) the standard deviation of the linearized signal for a 64 × 64 pixel patch. For colour cameras, weighted output of the R, G, B channels is used.
The standard again provides for the situation in which the focal plane exposure cannot be measured. The ISO standard (12232) includes weighting factors for summing luminances in single-exposure colour cameras in which the weighting factors are not known, from the individual channels according to ITU-R BT.709:
The standard deviation is then calculated using the weighted formula:
Standard output sensitivity (SOS)
To evaluate standard output sensitivity the sensor is exposed to produce a signal level of 461/1000 OMAX, where OMAX is the maximum output value of the system. SOS is then given by:
where HSOS is the exposure required. HSOS values are rounded using tables given in the standard before reporting and should give an indication as to the use of tungsten or daylight illumination.
Recommended Exposure Index (REI)
Recommended exposure index (REI) may be calculated using:
where HM in this instance is an exposure that the camera or sensor manufacturer specifies. This method of speed rating can serve to account for instances where mathematical processing or the exposure regime of the camera is unconventional and cannot be specified using the recommended methods, e.g. multiple-exposure high-dynamic-range cameras. It can further account for when the camera is to be used for specific circumstances such as document imaging. The value should not be used for comparison of systems.
Digital camera ISO settings
On a wide variety of digital cameras, it is possible to change the ISO settings at will and as such it appears superficially that manufacturers have discovered a method for arbitrarily changing the sensitivities and thus speeds of their systems. A typical digital single-lens reflex (SLR) camera may have an ISO range of 50–6400. In practice all cameras have a nominal speed as defined by the ISO standards described above and this will generally be the slowest speed represented in the cameras’ ISO range. When a user selects a higher ISO speed rating the camera sacrifices signal levels to allow the user to use shorter exposure times and subsequently adjusts the analogue and digital gains and image processing to produce the image. Therefore, as ISO speed ratings are increased, the noise levels in the image will rise. The facility is added for the convenience of users and purchasers should use the nominal speed rating of the camera for comparisons.
TYPICAL SENSITIVITIES OF FILM AND CCD DEVICES AND SPEED RATINGS IN PRACTICE
The only useful speed number in practice is one that adequately represents the speed of a material, or system, under the conditions in which it is likely to be used. Published speed values aim at providing exposures that will secure a printable negative, or an excellent result from a digital system under a wide range of conditions. The published rating for a given material or system cannot therefore be the best to use under all conditions of use. For critical use it is always best for the user to determine the appropriate exposure to suit the particular equipment and conditions. Published exposure indices provide a valuable starting point, although if you have selected an exposure index that suits your needs you should remember that it is no more ‘correct’ than the published figure, except for your own conditions of working. Typically, emulsions range from ISO 64 for fine-grained colour reversal, through ISO 100 for fine-grained colour and black-and-white print films, ISO 200–400 for medium-grained emulsions, to ISO 800 for fast colour print films and even ISO 1000 for some black-and-white materials. There are a number of black-and-white emulsions that are marketed to be exposed at speeds of up to 3200. This, however, is a result of push processing and their ISO rating is typically lower. Digital cameras have nominal ratings of between ISO 100 and 400 in the consumer market, though they may be exposed up to ISO 25600. Typically, noise levels in digital images exposed beyond ISO 800 start to become objectionable.
For further sensitivity data for specific photographic products and CCD cameras, the reader is recommended to consult the websites of the manufacturers.
BIBLIOGRAPHY
Allbright, G.S., 1991. Emulsion speed rating systems. Journal of Photographic Science 39, 95–99.
Dainty, J.C., Shaw, R., 1974. Image Science. Academic Press, New York, USA.
Egglestone, J., 1984. Sensitometry for Photographers. Focal Press, London, UK.
Holst, G.C., 1998. CCD Arrays, Cameras and Displays, second ed. SPIE Optical Engineering Press, Bellingham, WA, USA.
ISO 2721:1982, 1982. Photography – Cameras – Automatic Controls of Exposure.
ISO 7589:1984, 1984. Photography – Illuminants for Sensitometry – Specifications for Daylight and Incandescent Tungsten.
ISO 5800:1987, 1987. Photography – Colour Negative Films for Still Photography – Determination of ISO Speed.
ISO 6846:1992, 1992. Photography – Black-and-White Continuous Tone Papers – Determination of ISO Speed and ISO Range for Printing.
ISO 6:1993, 1993. Photography – Black-and-White Pictorial Still Camera Negative Films/Process Systems – Determination of ISO Speed.
ISO 2240:1994, 1994. Photography – Colour Reversal Camera Films – Determination of ISO Speed.
ISO 14524:1999, 1999. Photography – Electronic Still Picture Cameras – Methods for Measuring Optoelectronic Conversion Functions (OECFs).
ISO 12232:2006, 2006. Photography – Digital Still Cameras – Determination of Exposure Index,
ISO Speed Ratings, Standard Output Sensitivity and Recommended Exposure Index. ISO 17321:2006, 2006. Graphic Technology and Photography – Colour Characterisation of Still Cameras (DSCs).
ITU-R BT.709:1993, 1993. Basic Parameter Values for the HDTV Standard for the Studio and for International Programme Exchange.
Jacobson, R.E., Ray, S.F., Attridge, G.G., Axford, N.R., 2000. The Manual of Photography, ninth ed. Focal Press, Oxford, UK.
James, T.H. (Ed.), 1977. The Theory of the Photographic Process, fourth ed. Macmillan, New York, USA.
Kriss, M.A., 1998. A model for equivalency ISO CCD camera speeds. Proceedings ICP5 ’98, Antwerp, Belgium, Volume 2, pp. 341–347.
Nakamura, J. (Ed.), 2006. Image Sensors and Signal Processing for Digital Still Cameras. Taylor & Francis, New York, USA.
Todd, H.N., Zakia, R.D., 1974. Photographic Sensitometry. Morgan & Morgan, New York, USA.
Walls, H.J., Attridge, G.G., 1977. Basic Photo Science. Focal Press, London, UK.