Imaging the Invisible - Using Modified Digital Still Cameras for Straightforward and Low-Cost Archaeological Near-Infrared Photography more

Journal of Archaeological Science 35 (2008) 3087–3100 Contents lists available at ScienceDirect Journal of Archaeological Science journal homepage: http://www.elsevier.com/locate/jas Imaging the invisible using modified digital still cameras for straightforward and low-cost archaeological near-infrared photography Geert Verhoeven* Ghent University, Faculty of Arts and Philosophy, Department of Archaeology and Ancient History of Europe, Blandijnberg 2, B-9000 Ghent, Belgium a r t i c l e i n f o Article history: Received 11 February 2008 Received in revised form 2 June 2008 Accepted 6 June 2008 Keywords: Near-infrared Digital NIR photography Digital still camera Camera modification NIR filter a b s t r a c t Analogue near-infrared (NIR) photography has already been used a lot in both scientific and medical photography, in which cases near-infrared (NIR) radiation was mostly captured by InfraRed (IR) sensitive plates or film emulsions. However, its use in archaeology has remained rather restricted, most likely due to some ignorance and/or lack of knowledge about this kind of photography, while the critical imaging process also severely limited its use. This situation could be, however, changed completely, as the image sensors used in digital still cameras (DSCs) are very sensitive to NIR wavelengths, making the quite lengthy and error-prone film-based NIR imaging process obsolete. Moreover, modifying off-the-shelf DSCs even simplifies this digital acquisition of NIR photographs to a very large extent. By starting with a general outline of the ElectroMagnetic (EM) spectrum and the specificities of NIR radiation, the base is laid out to tackle the possibilities and practicalities of archaeological NIR imaging, subsequently comparing the earlier film-based approach with the digital way of NIR shooting, showing how the latter can greatly benefit from modified compact, hybrid and small-format Single Lens Reflex (SLR) DSCs. Besides in-depth information on the technique of digital NIR photography, examples will illustrate its archaeological potential. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction Since the astronomer and composer Sir Frederick William Herschel (1738–1822) discovered in 1800 the InfraRed (IR) portion of the ElectroMagnetic (EM) spectrum, a lot of scientific disciplines have become fascinated by this kind of invisible radiation. This interest even increased after World War II, as Colour InfraRed (CIR) emulsions had shown their capabilities. These days, IR photography is used a lot in forensics, astronomy, aerial survey, (bio)medical and several other types of scientific photography, although its use in archaeology is rather disappointing. Because this can largely be attributed to a lack of knowledge about the technical specification of near-infrared (NIR) radiation and the concept of EM radiation, the specific character of NIR will first be explained before tackling the possibilities and practicalities of archaeological NIR imaging. 1.1. EM radiation and NIR The light that makes the Human Visual System (HVS) perceive the world around us is in fact an EM wave, comprising two oscillating magnetic and electric fields perpendicular to each other as * Tel.: þ32 9 264 41 39; fax: þ32 9 264 41 73. E-mail address: geert.verhoeven@ugent.be 0305-4403/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jas.2008.06.012 well as perpendicular to the direction of propagation (Slater and Frank, 1974; Waldman, 2002) and travelling at 299792.458 km/s in vacuum, a speed that decreases when light travels in air, glass, water or other transparent substances (Young and Freedman, 2004). This explanation, put forward almost two centennials ago by the Scottish physicist James Clerk Maxwell (1831–1879), was based on the findings of the Dutchman Christiaan Huygens (1629–1695), who already declared light to travel in the form of waves (Clegg, 2001; Waldman, 2002). Being a wave phenomenon, the wavelength (l) is the most important characteristic of EM radiation. The EM waves humans perceive – the so-called visible light – encompasses a very small portion of all EM radiation: only wavelengths between approximately 380 nm and 750 nm (Fig. 1), the absolute thresholds varying from person to person and specific viewing conditions. However, on both sides of this extremely small visible spectrum resides EM radiation the HVS is insensitive for, characterised by wavelengths smaller than 380 nm or larger than 750 nm. Just as visible light, these wavebands were divided into spectral regions and given names as gamma rays, X-rays and UltraViolet (UV) rays on the short-wavelength side, while IR rays, microwaves and radio waves can be found in the long-wavelength region (Fig. 1). The aforementioned IR portion of the EM spectrum comprises wavelengths between 750 nm and 1 mm, hence spanning three orders of magnitude. As the width of this particular band is 3088 G. Verhoeven / Journal of Archaeological Science 35 (2008) 3087–3100 (Richards, 2001), energy given off by all real-world objects (Barnes, 1963). As a matter of fact, all objects with a temperature above absolute zero (0 K or À273.15  C) emit EM radiation, but the type and amount of the latter largely depends upon the temperature of the matter. A healthy human body with a temperature of 310 K (circa 37  C) gives off wavelengths with a peak around 9350 nm (LWIR) and no detectable amounts of NIR energy. With a rising temperature, the amount of radiated EM radiation will increase and the wavelength of maximum emittance lmax will be shorter (as described by Wien’s displacement law). Practically, objects need to be heated to about 500 K before they start radiating in the NIR range, while a temperature of at least 800 K (e.g. an electric stove burner) must be attained before visible red light is emitted (Barnes, 1963; Ray, 1999). Because it is not possible to photograph IR radiation but the NIR portion with conventional film-based approaches or digital photo cameras, performing NIR photography basically boils down to capturing the particular amounts of reflected NIR radiation emitted by very hot objects such as the sun, incandescent light bulbs or specific extraneous NIR sources, rather than recording the ambient temperature variation. This misunderstanding is also fed by some of the older (archaeological) literature. Simmons (1969, p. 94) for instance literally says: ‘‘cool objects appear dark, warm objects appear light; hence green vegetation looks very whitish, while water is blackish’’. Although water indeed appears black on an NIR photograph, this response has nothing to do with its temperature, but is due to a very high NIR absorption, indicated by its absorption coefficient (Curcio and Petty, 1951). 1.3. Two techniques Fig. 1. The ElectroMagnetic spectrum (adapted from Freedman and Kaufman, 2005, Figs. 5–7). exceptional, it is often subdivided into several zones, for which the limits (and also the number of subdivisions) are to a certain extent dependent on discipline and largely varying through literature. In general, the following breakdown can be used (based on Daniels, ¨ 2007; Deutsches Institut fur Normung, 1984): 1. 2. 3. 4. 5. Near-infrared (NIR) from 750 nm to 1400 nm (1.4 mm); Short Wavelength InfraRed (SWIR) from 1.4 mm to 3 mm; Mid Wavelength InfraRed (MWIR) from 3 mm to 6 mm; Long Wavelength InfraRed (LWIR) from 6 mm to 15 mm; Far/Extreme-InfraRed (FIR) from 15 mm to 1000 mm. Moreover, EM radiation can also be thought of as a travelling bundle of particles. Although this theory was contested by many since its launch by the Englishman Isaac Newton (1642–1727), Albert Einstein (1879–1955) finally demonstrated the existence of such discrete energy packets, now called photons (Clegg, 2001; Waldman, 2002). This wave-particle duality is still one of the key concepts in quantum mechanics, signifying EM radiation exhibits both wave and particle behaviours. Depending on the wavelength of the EM radiation, the energy of the photons differs. Applied to NIR, it means this type of radiation contains less energetic photons compared to visible light. In NIR imaging, two techniques exist. The first, and by far most applied, kind of photography uses the already mentioned reflected/ transmitted portion of the incident NIR radiation. Every object exposed to NIR radiation will absorb, reflect and transmit these incident photons to some extent. Recording these particular amounts is generally termed reflected (N)IR photography and should not be confused with IR reflectography, the latter using longer wavelengths up until around 2000 nm (van Asperen de Boer, 1966, 1969). Secondly, there is NIR fluorescence photography (sometimes called NIR luminescence – Barnes, 1963; Bridgman and Gibson, 1963; Gibson, 1962, 1963a,b), in which the NIR sensitive medium records in fact fluorescence, being radiation emitted by the subject under study in the NIR region. Rather than directly being emitted as in the case of very hot objects, these NIR photons are excited upon being exposed to incident shorter wavelengths (mostly UV or visible blue and green wavelengths). Although its application is mainly restricted to the forensic field, the very strong fluorescence of particular minerals and pigments (Barnes, 1958) makes this type of NIR photography worthwhile in certain archaeological case studies. 2. NIR imaging in archaeology It lasted till the 1930s for NIR sensitive emulsions to become relatively available, allowing photographers to practise this new technique with a certain ease and certainty. From this period onwards, the possibility to visualise an often subtle, dissimilar behaviour of materials in the NIR helped archaeologists to depict certain object characteristics not (or less) apparent to the HVS. However, despite its application and the great deal of work done in a wide variety of scientific research fields, the scale of utilization always remained very small from an archaeological perspective – for instance illustrated by the fact that Cookson’s ‘‘Photography for Archaeologists’’ (Cookson, 1954) does not mention one single time the use of NIR. In broad terms, archaeologically inspired NIR 1.2. NIR imaging There exists a lot of confusion and misconception about NIR imaging, often linked with images like the ones displayed in Fig. 2. However, these so-called heat images were yielded by electronic thermography, a technique based on a completely different part of the EM spectrum. Heat imaging uses the MWIR and LWIR G. Verhoeven / Journal of Archaeological Science 35 (2008) 3087–3100 3089 Fig. 2. (A) LWIR image of a man’s face and (B) MWIR image of a four-fingered hand (Richards, 2001, Figs. 6.2 and 2.19). photography has been applied over the years to answer questions in all sorts of research areas. 2.1. The investigation and decipherment of documents Charred, dirty, worn, bleached, censured, obliterated, faded or very deteriorated documents often reveal their secrets through recording their NIR reflectance and/or fluorescence. While some inks are largely transparent, others do reflect NIR to a larger extent. Moreover, NIR often can differentiate between dyes and pigments ¨ that look indistinguishable to the HVS (Bruning, 1939; Ross, 1933). Exploiting these spectral signature properties might bring document alterations to light or reveal the underwriting of obliterated passages in case the top ink is less opaque to NIR than the ink below. This way, direct NIR photography made a faint black Egyptian text from about 1200 B.C. written on very dark brown and aged leather sufficiently readable again (L.D., 1933), while other scholars deciphered original writings on fragments of badly discoloured (Beardsley, 1936) or charred papyri (Chabries et al., 2003). Famous is also the work of the Arab photographer Najib Anton Albina, who used reflected NIR photography in the 1950s to retrieve information from the Dead Sea scroll fragments (Schiffman, 1995), whereas UV induced NIR fluorescence disclosed some of Vindolanda’s wooden Roman writing tablets (Rutherford, 1977). The results achieved by Coremans (1938) even proved NIR photography to reveal writings and drawings underneath encrustations (Fig. 3) or a patina. 2.2. The examination of textiles and tapestry fabrics, even if they appear visually similar (Dorrell, 1994). This way, Coremans (1938) detected tapestry restoration. Baldia and Jakes (2007) even incorporated NIR photography in a complete range of photographic methods for non-destructive research of archaeological textiles. 2.3. Inspection of paintings The same principles hold for the investigation of all kinds of paintings: on canvas, rock, wood or glass. Being a true pioneer, Marshack (1975) utilised NIR photography in some French caves to get a deeper insight into the employed pigment and the dating of the Cro-Magnon man’s cave art, while NIR imaging also proved several times to be indispensable in the detection of painted forgeries. Although such research is situated often in the field of art studies, proving the authenticity and examining the fake or altered state of the canvas also largely rely upon NIR fluorescence (Bridgman and Gibson, 1963) and direct NIR photography (Aldrovandi et al., 1988; Lyon, 1934), as the latter might reveal information which the technique of IR reflectography is unable to unveil (Gargano et al., 2007). 2.4. Tattooing research In several studies, NIR reflected radiation has been taken into account to detect forms of tattooing on human remains that were naturally or artificially mummified (Alvrus et al., 2001; Smith and Zimmerman, 1975). 2.5. Pottery study By differentiating between the NIR reflectance and/or transmittance of the particular pigments, dyes and materials used, NIR photography also furnishes a useful means for the examination of Traces of pigments can also be found on ceramics, in which case photography by NIR can help enormously in the study of tituli picti, Fig. 3. (A) Panchromatic and (B) reflected NIR photograph of the same artefact (adapted from Coremans, 1938, Figs. 13a and b). 3090 G. Verhoeven / Journal of Archaeological Science 35 (2008) 3087–3100 decoration, etc. Used in combination with radiography, Milanesi (1963) demonstrated the benefit of NIR photography in dating and classifying prehistoric and protohistoric pottery fragments. 2.6. Aerial photography Notwithstanding its rather large potential in aerial archaeology, very little non-visual photography has been executed, although it still remains one of the fields where most archaeological NIR-based research is undertaken (e.g. Agache, 1968; Braasch, 2007; Edeine, 1956; Gumerman and Neely, 1972; Hampton, 1974; Powlesland et al., 1997; Rigaud and Bouyer, 1986; Strandberg, 1967; Verhoeven, 2007). As healthy vegetation appears very bright and the NIR reflection of a stressed, diseased or dead vegetation canopy is much less abundant, cropmarks can often be distinguished better in this part of the EM spectrum (Fig. 4; for a more in-depth discussion, consider Verhoeven et al., submitted for publication; Verhoeven, 2007). Besides, reflected NIR photography enhances certain soilmarks because water largely absorbs incident NIR, while the same principle makes it very easy to discern bodies of water, the latter generally appearing black on the photograph. Thirdly, the long NIR wavelengths largely penetrate the haze often present when imaging landscapes, especially in the case of high altitude and oblique aerial photography, hence increasing the visibility of distant objects (Conlon, 1973; Ross, 1933). The explanation lies in the fact that atmospheric scattering is much less in the NIR region than it is in the visible part, yielding imagery with enhanced clarity of detail and a larger contrast (Rawling, 1946). This short overview indicates that NIR photography certainly holds several benefits in archaeological research, while it is most likely that much more subjects are very suited for this kind of imaging (Conlon, 1973), or as Matthews says: ‘‘It so often happens that all other methods failing to record the desired result, that the operator, often in desperation, experiments – and finds his answer in photography by infra-red’’ (Matthews, 1968, pp. 131–132). Till only very recently, archaeological NIR photography was performed using films whose emulsion was sensitised into the NIR range. Even though recent advantages in digital technology made this cumberstone process obsolete, it is best to review how NIR imaging was performed in the film era before expatiating on the digital approach and its additional advantages. 3. NIR imaging in the film era Although generally the same cameras and light sources can be used as for imaging reflected visible light, NIR photography still features some peculiarities. In the following overview, the major changes and additional requirements over ‘‘normal’’ photography will be treated (here to be considered small format/35 mm frame photography). 3.1. Emulsions IR sensitive film exists in two variants, being Black-and-White negative NIR film (monochromatic IR film) such as Kodak Professional High-Speed Infrared Film (HIE), or the (False-)Colour Infrared (FCIR or CIR) film, from which the Kodak Ektachrome Professional Infrared Film EIR film is an example. As it also features only one layer of emulsion, the first film type is to a certain extent similar to a conventional panchromatic emulsion, although it is sensitised up to 900 nm. Due to its non-uniformity in the visible bands, this kind of film is often used for true NIR photography: capturing only wavelengths between 750 nm and 900 nm by fitting an appropriate filter on the lens. (F)CIR film, which comes in both negative/print and positive/ reversal/transparency alternatives, also bears great resemblance in construction to colour film, being an integral tripack material which makes use of dyes added to the silver, spectrally sensitizing the silver-halide grains. However, the three individual emulsion layers react with different portions of the EM spectrum. The upper layer, which contains cyan dyes, only reacts to NIR, while the other two layers, containing Yellow and Magenta, react to Green and Red light, respectively (Fig. 5) (Eastman Kodak Company, 1968; Kodak´ Pathe, 1979). As all three layers also respond to Blue radiation, a Yellow (minus-Blue) filter is used (or implemented in the emulsion) to cut out its image degrading effect, besides limiting the sensitivity of the layer to its intended spectral region. By the effects of exposure and processing, each individual layer produces a dye of a complementary colour (Fig. 5). Consequently, NIR wavelengths are imaged as Red, while reflected Green radiation is visualised as Blue and Red reproduces as Green. This way, one subtractive colour (Cyan, Magenta or Yellow) controls each primary colour, while the colours of the original object are remapped to pseudo-colours, attributing to this kind of film the term false-colour. No matter which type of film is used, the NIR sensitivity of most films dramatically drops around 900–925 nm (their specific upper cutoff wavelength depending upon the dyes added), with an emulsion as Konica Infrared 750 even responding to only 820 nm (Konica s.d.). 3.2. Lenses 2.7. Excavation photography Both in imaging the horizontal and vertical sections, NIR site photography has been proved very useful. Some examples are the work of Buettner-Janusch (1954), who claimed features and stratifications to be visible in pure NIR imagery, while not discernable on the panchromatic frames (i.e. Black-and-White film sensitive to all visible wavelengths) or to his eyes. The same conclusion was drawn by Reichstein (1974), who effectively acquired Black-and-White NIR imagery to reveal previously indiscernible Bronze Age plough marks. Additionally, Hirsch (1975) definitely demonstrated the presence of painted floors at the Cretan Archaeological site of Gournia, as NIR photography revealed the colour of one of the floors to be owed to pigment rather than clay. Fig. 4. Kodak EIR image of a dense archaeological landscape containing Neolithic and Roman features (Braasch, 2007, Fig. 8). Just as in ‘‘normal’’ photography, about any lens can be used in ´ NIR photography (Kodak-Pathe, 1979) as the majority of optical G. Verhoeven / Journal of Archaeological Science 35 (2008) 3087–3100 3091 Fig. 5. Colour reproduction with a CIR transparency film (adapted from Eastman Kodak Company, 1997, Fig. 2). Fig. 7. The lay-out and working of a photodiode array with Bayer CFA. glasses and polymers freely transmit NIR (Ray, 2002), although the design of the lens still counts when one wants to avoid a hot spot: a brighter area in the centre of the image, produced by internal light reflections caused by the lens coatings. Moreover, because NIR radiation features longer wavelengths than visible light, the waves are less refracted and focus to another point which lies behind the visible focal plane (a phenomenon called longitudinal chromatic aberration – Hackforth, 1960). To counteract, the lens has to be focused on an object that is closer than the actual object, a process known as short focusing. This explains why many pro and/or older lenses have a red dot/line/letter ‘‘R’’ indicating the NIR focus offset when focused at infinity (Nieuwenhuis, 1991). Newer lenses seldom display such a supplementary focusing index, nor do they have an NIR Depth Of Field (DOF) scale (the latter indicating which parts of the subject, expressed as a distance in meter or feet from the lens, will be rendered sharp). Besides short focusing, a smaller diaphragm setting (e.g. f/8 or f/11) – with corresponding extended DOF – will largely solve most of these focus errors. Too small an aperture may be counterproductive, as diffraction effects start to come into play and shutter speeds might become lengthy. 3.3. Filters As all aforementioned films are largely sensitive to visible light, optical filters are needed to exclude parts of this spectrum (or completely in case of pure NIR photography). In a typical photographic situation, a deep red filter (most commonly a Wratten 25, B þ W 092 or similar) is attached to the front of the lens, only allowing some Red and NIR radiation to pass. Such filters make sure the photographer can still see a dim image in the viewfinder, enabling him/her to compose and focus, while not completely loose the NIR look (Gibson, 1968, 1978). In case a pure NIR image is required, an IR filter is needed. These completely visually opaque filters (type Hoya R72, Wratten 87 or similar) transmit only radiation in the NIR waveband (Conlon, 1973), hence prohibiting visible light from being recorded (the cuton wavelength and exact amount of radiation transmitted being filter dependent). As composing and focusing is now made completely impossible, the lens must be prefocused and the image composed before installing the filter on the lens. This inconvenient situation even becomes worse when using lenses with different threads, as this necessitates filters in different sizes. In studio situations, this could largely be resolved by placing the appropriate filter over the light source instead. As different filter brands exist (B þ W, Cokin, Hoya, Kodak, Lee, Schott) all with their specific screw-in or gel type filters, it is advisable to check the particular spectral curves or transmittance chart for each filter, as it is impossible to use only one, ideal filter for all purposes. The fact that incident radiation generates longer wavelengths constrains two different filters to successful accomplishment of NIR fluorescence photography: first of all, an appropriate exciter/excitation filter must be placed over the light source (generally a blue/ green-pass filter) to make the latter only emit shortwave visible light and prevent any NIR photons to hit the subject. Secondly, the aforementioned opaque NIR filter (in this case called barrier filter) on the lens must make sure only the NIR excited photons are captured, excluding the reflected visible light (Gibson, 1962). 3.4. Disadvantages This film-based approach has, however, some major disadvantages, which make the complete workflow rather critical and NIR imaging difficult to master. 3.4.1. Exposure Being generally the biggest obstacle to getting a good NIR photograph, a solid knowledge of NIR principles is needed to make the right decisions about exposure. The problem lies in the fact that normal exposure meters do not take NIR into account. Consequently, the exposure reading one gets is based on the visible light hitting the exposure sensor rather than the NIR radiant flux. Moreover, NIR sensitive films only have a recommended kind of Fig. 6. Spectral sensitivity characteristics of Sony’s ICX205AL CCD (adapted from Sony Corporation s.d.) and a Kodak CMOS Image Sensor (adapted from Eastman Kodak Company, 2003, Fig. 1). 3092 G. Verhoeven / Journal of Archaeological Science 35 (2008) 3087–3100 Fig. 8. Wavelength versus Quantum Efficiency for (A) the Kodak KAF-8300 and (B) the Foveon X3 sensor (adapted from Eastman Kodak Company, 2005a, Fig. 5; Lyon and Hubel, 2002, Fig. 6). film speed instead of a fixed ISO value, as the latter is always defined using the visible part of the EM spectrum (Eastman Kodak Company, 2005b). Consequently, the exposure reading has to be considered a starting point, after which over- or underexposure is needed according to the subject and the lighting conditions photographed in (time of the day, time of the year, geographical location, weather conditions, NIR source). Only with enough experience, the photographer can estimate the quantity of NIR radiation reflected from or emitted by particular objects. Certainly in aerial work, this is a big disadvantage, as the amount of NIR strongly varies from scene to scene. Consequently, bracketing – i.e. taking a series of photographs with different exposures – is advised. 3.4.2. Sensitivity Notwithstanding the improvements made, the current emulsions are only sufficiently sensitive to be used in handheld cameras with rather large amounts of reflected NIR radiation. Moreover, the application of filters additionally extends the exposure time, meaning aerial photography is for instance impossible but in very sunny circumstances (Hampton, 1974), as other conditions would considerably increase the risk to acquire blurred imagery. 3.4.3. Storage, handling and processing Although the films cannot detect thermal radiation, their extended sensitivity makes them more susceptible to fogging. NIR emulsions therefore have to be stored and transported refrigerated (Eastman Kodak Company, 1997). Storage in less than ideal conditions can easily lead to sensitometric changes. Not only opening the black plastic film canister, but also loading and unloading the camera have to be executed in absolute darkness (Gibson, 1968). After exposure, the film must be developed as soon as possible, but not without the necessary precautions: e.g. using NIR-safe processing tanks, turning off any IR sensor in the labs, etc. All these elements made NIR imaging a rather tricky business, and although experienced photographers certainly could get very decent results, the final outcome could never be entirely predicted. Consequently, NIR (archaeological) photography was usually only attempted by skilled photographers, scientists, and technicians with a particular purpose in mind. Add to this the cost of the NIR film, one can already imagine the huge advantage and democratization a digital approach could mean to NIR photography. In spite of this, the use of NIR imaging with digital consumer or SLR cameras was till now never fully explored in archaeology. 4. Digital-based NIR imaging Since the 1990s, much has changed in the photographic world. Certainly since the advent of the 21st century, there is an ever increasing growth of digital shooters due to the large availability of sophisticated but affordable digital still cameras (DSCs) and major advances in computer technology. Unlike video or silver halide photographic cameras, a DSC equals a camera equipped with both a digital image sensor for capturing photographs and a storage device for saving the obtained image signals in a digital way (Toyoda, 2006). Besides all the well-known benefits in acquisition, storing, manipulating and retrieving digital imagery, DSCs offer an additional major advantage, making NIR photography much less awkward. Fig. 9. Spectral sensitivities of (A) the Nikon D80 and (B) the Foveon X3 sensor with CM500 NIR-cut filter (adapted from Image Engineering s.d; Lyon and Hubel, 2002 Fig. 7). G. Verhoeven / Journal of Archaeological Science 35 (2008) 3087–3100 3093 although the spectral sensitivity and response increased to a very large extent when compared to photographic film. 4.2. Filters and microlenses Although it is the heart of every digital camera, it is not the intrinsic NIR response of the imaging sensor alone that determines the final spectral response of the DSC. Whatever the type, all image sensors consist of a 2D-photosite array to generate a digital photograph. The light-sensitive area, called photodiode, collects the photons during the exposure time (Holst, 1996; Janesick, 2001; Nakamura, 2006; Theuwissen, 1995). A digital camera which yields 12.2 million pixels has a sensor built-up by at least 12.2 photodiodes (e.g. 4288 columns  2848 rows), as one photodiode generally contributes one effective image pixel. The percentage of the photosite that is sensitive to the incoming light is called the fill factor (FF). Improving this FF is achieved by an array of on-chip microlenses, collimating incident photons to the photodiode (Dierks, 2004; Holst, 1996; Nakamura, 2006; Theuwissen, 1995). Below this array of small lenses, an on-chip Colour Filter Array (CFA) is positioned. This mosaic pattern of thin, coloured filters is positioned on top of all but the Foveon sensor (which captures all three basic colours at every location), each optical filter transmitting only specific wavelengths of the incident radiant energy (Holst, 1996; Nakamura, 2006). Every individual photosite of the array is covered by such a filter, allowing only one – but broad – particular range of wavelengths to be transmitted and subsequently captured by the photodiode (Fig. 7). Generally, small-format DSCs use a Red–Green–Blue (RGB) pattern with a repeating group of four photodiodes, in which two have Green filters and the remaining are either Red or Blue (Fig. 7). This arrangement is called a Bayer pattern, after Bryce Bayer of the Eastman Kodak Company who patented it in 1976. The occurrence of twice as many Green filters as Blue or Red filters mimics the HVS, whose spatial resolving power for luminance information is larger than for chrominance data (Bayer, 1976). As this critical luminance information is mostly carried by Green light, the higher occurrence of Green filters improves the spatial sampling of the luminance signal and hence the perceived sharpness of the digital image (Bayer, 1976; Parulski and Spaulding, 2003). The specific characteristics of both CFA and microlens array (e.g. their thickness and opacity to particular wavelengths) make the final response to NIR (or to EM radiation in general) varying from camera to camera according to the used matrix, but it can be stated that DSCs are in general very sensitive to NIR radiation. This is confirmed by the examples shown in Fig. 8A and B. The former depicts the response of the photodiodes covered with Red (R), Blue (B) and the two types of Green glasses (GRr and GBr) of the KAF-8300 Kodak CCD sensor, previously being used in the Olympus E-300 and E-500. In Fig. 8B, the wavelength-dependent Quantum Efficiency (QE) for the Foveon X3 sensor is plotted as a function of wavelength. Note that this particular kind of CMOS sensor, which is incorporated in Sigma’s DSLRs (SD9, SD10 and SD14) as well as in Polaroid’s x530 point-andshoot camera (Foveon, 2007), does not use a CFA. Because of its layered design, this imager measures every visible colour at every photosite (Foveon, 2007; Lyon and Hubel, 2002). 4.3. Low-pass and NIR-cut filter The spectral curves in Fig. 8A and B are, however, not the result one would get when determining the spectral response of all three channels in the final manufactured DSC. To cut down the imagequality-degrading NIR and enhance the true colour rendition, camera manufacturers place an NIR-blocking/cutoff/cut filter (also called hot mirror) in front of the whole imaging array. By reflecting and/or absorbing all NIR radiation that passes the lens (Busch, 2007; Koyoma, 2006; Ray, 2002), these filters allow only photons Fig. 10. Modification of the digital imaging array by replacing the hot mirror. 4.1. Sensor Being it the often used Charge-Coupled Device (CCD) and Complementary Metal Oxide Semiconductor (CMOS) or the less implemented Junction Field Effect Transistor (JFET) and CMOS-FoveonÔ X3 sensors, pure silicon (Si, atomic number 14) is the basis for essentially all digital image sensors of current DSCs: from low-cost consumer point-and-shoot and hybrid cameras to expensive, professional small-format D-SLR (Digital Single Lens Reflex) cameras. Pure silicon is a semiconductor: a material whose physical properties lie somewhere in-between electricity conductors and insulators. It exists in the form of a crystal lattice with each silicon atom bonded covalently to four other silicon atoms (Silberberg, 2006). Because this arrangement does not allow silicon to conduct electricity very well, small amounts of impurity are introduced into digital image sensors to increase the silicon’s conductivity – a process called doping (Holst, 1996; Rieke et al., 1994; Silberberg, 2006; Theuwissen, 1995). When light enters the sensor, photons impinging on and penetrating into the silicon crystal can promote electrons to a higher energy state as long as their energy level is higher than the so-called bandgap (i.e. the energy difference between the conduction band and the valence band) of silicon, which is approximately 1.12 eV at room temperature (Holst, 1996; Janesick, 2001; Rieke et al., 1994; Theuwissen, 1995). This means that any photon of radiant energy till about 1127 nm (the so-called cutoff wavelength lc) has enough energy to promote an electron from the valence band to the conduction band in crystalline silicon, allowing for the photon-to-signal charge conversion to take place (Nakamura, 2006; van de Wiele, 1976). This physical fact precludes normal digital sensors for detection of wavelengths beyond about 1100 nm, but implies digital cameras to have high Quantum Efficiency (QE) in the shorter wavelengthpart of NIR (Mullikin et al., 1994). Since four broad types of image sensors exist and different manufacturing processes are used, the general spectral response of digital image sensors tends to vary (see Fig. 6 for the typical response of two monochrome sensors), 3094 G. Verhoeven / Journal of Archaeological Science 35 (2008) 3087–3100 Fig. 11. Contrast enhancement (A2 and B2) on a visible (A1) and an NIR frame (B1) clearly shows the latter to reveal better Roman features as cropmarks. from the intended part of the visible spectrum to be transmitted. The resulting spectral response range of the complete image array to incident EM radiation is given in both Fig. 9A and B, again showing the occurring varieties. Although it depends on the quality of the NIR-blocking filter, most DSCs coming straight out of the box can still be used for NIR imaging, as most filters still transmit a few percentages NIR radiation, hence allowing a small portion of the incident NIR wavelengths to reach the image sensor. Besides a visibly opaque NIR filter in front of the lens (as in the film-based approach), an additional tripod is still needed, because the very low NIR sensitivity inevitable compels very long shutter speeds. This excludes the use of these cameras in research where accurately and fast composing as well as short shutter speeds is crucial (e.g. aerial photography). 4.4. Modification Although different researchers already used off-the-shelf professional and/or consumer DSCs for pure or false-colour NIR imaging (e.g. Milton, 2002; Fredlund and Sundstrom, 2007), replacing the internal hot mirror of these cameras with a visibly opaque filter (Fig. 10) makes NIR imaging much easier and more predictable, while G. Verhoeven / Journal of Archaeological Science 35 (2008) 3087–3100 3095 Fig. 12. A conventional (A1) and FCIR record (B1) of the central part of the Roman city Trea. The enhanced versions (A2 and B2) point to the fact that FCIR can be advantageous in spotting cropmarks, clearly illustrated by the walls of the Roman temple in the insets. hugely increasing the camera’s NIR sensitivity. In fact, the filter that is placed in front of the sensor can be chosen according to the specific needs to capture particular wavebands. Using a clear filter enables the whole spectral sensitivity range of the camera (from UV to NIR) to be used, while inserting an NIR-pass/cold filter will allow pure NIR imaging, as the final image will be formed by NIR radiation only. In the end, this operation boils down to moving the previously lensfitted filter to the sensor, but with the additional advantages of large NIR responsivity (which is much larger and more extended compared to film) and the possibility to look through the camera’s viewfinder. However, one must take into consideration that clear (or other visible transmitting filters) still require an opaque filter to be fitted onto the lens in the case of pure NIR photography. Replacing this internal IR-block filter is, however, a delicate job, as high voltages are present inside the camera and a lot of dust can stick to the sensor. To exclude the risk of damaging the camera (and losing the money spent as conversion voids the manufacturer’s warranty), it is advisable to have this modification performed by a dedicated company (e.g. Khromagery, LDP LLC, Life Pixel). Although these companies often modify only a few particular DSCs (mostly D-SLRs), about any camera can have its NIR-block filter replaced. Finally, it remains important to mention that modification ´ makes DSCs somehow more subject to moire effects, as most cameras – certainly D-SLRs – combine the NIR-block filter with an optical anti-aliasing/low-pass filter, the latter slightly blurring the incoming signal as to generate less aliasing artefacts induced by the image sensor’s sampling. On the other hand, the removal of the low-pass filter results in pictures with increased sharpness in case no aliasing-prone subjects are shot (everything but meshes, fabrics,.). Fujifilm already has discovered these inherent possibilities of digital cameras back in August 2006, when they released the FinePix S3 Pro UVIR, a modified version of the Finepix S3 Pro in which the sensor is only covered with a clear glass, allowing for UV (partly), colour and NIR photography (Fujifilm, 2006). Besides not being sold in all countries, opaque filters are still required for pure NIR imaging, hence losing some of the modified camera’s major advantages such as automatic focusing and through-the-lens view (the latter partly resolved by incorporating live previewing on the rear LCD of the camera). Even though a full warranty comes with all their models (Fujifilm also released the hybrid IS-1 and D-SLR IS Pro in January and July 2007, respectively), the D-SLRs feature only full functionality as long as D and G type AF Nikkor lenses are used, making older lenses less and those from other brands almost completely unusable. 4.5. Lenses Using a dedicated modified D-SLR, the same lenses used for the film approach can be applied, although it is still best to test whether 3096 G. Verhoeven / Journal of Archaeological Science 35 (2008) 3087–3100 Fig. 13. The haze penetrating capabilities of the longer NIR wavelengths (B) compared to the visible record of a mountainous area (A). hot spots are generated by a specific camera-lens combination (Rørslett, 2004, 2007). The chance of creating such a hot spot is, however, much larger when utilising unmodified cameras, as the hot mirror will reflect the radiation that needs to be captured. This NIR radiation is subsequently reflected into the back element of the lens, where the occurring internal reflections largely contribute to flare and the creation of the visually unattractive and scientifically unjustifiable hot spot. 5. Real-world examples To explore the capabilities of NIR photography in aerial archaeology, a Nikon D50 was acquired and subsequently converted (for an in-depth discussion, please consult Verhoeven et al., submitted for publication; Verhoeven, 2007). This modification proved very useful in the aerial experiments executed so far. Fig. 11 shows two versions of the same scene: the eastern part of the Roman town Trea (43 180 4000 N, 13 180 420 W – WGS84), to be situated in Central Adriatic Italy (Regione Marche). The upper left image (A1) is a pure visible record created by an unmodified Nikon D200, while the converted Nikon D50 – coupled with the D200 on one camera rig and using an identical focal length – generated photograph B1 at the very same moment. A2 and B2 again show the same images, but now after their central portions have been subjected to some image processing (histogram stretching and local contrast enhancement) to expose the cropmarks more clearly. G. Verhoeven / Journal of Archaeological Science 35 (2008) 3087–3100 3097 Fig. 15. Comparison between the visible (A) and NIR record (B) of a coated rim fragment belonging to a Roman handmade pot (sherd by courtesy of Wim De Clercq – Ghent University). Fig. 14. Mutual comparison between visible (A) and NIR image (B) of a North African Late Punic/Early Roman bowl. (1) and (2) indicate zones of increased reduction (sherd by courtesy of Karen Ryckbosch – Ghent University). Directly comparing these two enhanced versions obviously shows the NIR record to reveal more cropmarks. This effect can largely be attributed to the fact that diseased, senescent and heavily nutrient/ water deficient vegetation is generally characterised by a large decrease in NIR reflectance (Carter and Estep, 2002; Lelong et al., 1998; Murtha, 1978; Wiegand et al., 1972). Very often, this NIR decrease is far more noticeable with respect to the visible band, which witnesses a global increase in reflected EM radiation. Consequently, NIR imagery often has a larger potential to depict (negative) cropmarks compared to conventional photographs. The second aerial example (Fig. 12) reveals the possibility to recreate the appearance of a FCIR film by using a two-camera system based on a converted DSC coupled with an unmodified DSC, and combining the pure NIR image yielded by the former with the Red and Green channel created by the conventional camera. To perform the latter operation, setting the DSC to save a RAW file (instead of a JPEG – Joint Photographic Experts Group – or TIFF – Tagged Image File Format) is advisable, as RAW enables full control over the originally captured Digital Numbers (DNs), hence making it the only valid choice for scientific photography (Verhoeven, submitted for publication). In this example, the visible and NIR RAW files – again captured above the aforementioned Roman city Trea – were converted in a completely linear way and without any White Balance (WB) applied, using a free ANSI C RAW decoder written by David Coffin: dcraw (Coffin, 2007). Afterwards, the images were saved as 16-bit TIFFs and their colour channels split. A pure NIR channel of the modified Nikon D50 was attributed to the Red channel of this new image, and merged with the Red and Green colour channels of the visible photograph (the latter two placed in, respectively, the Green and Blue channels of the new frame). The result (Fig. 12B1 and B2) is a digital FCIR image that is rendered the same way as in the analogue approach, but with a much better spectral fidelity. As can be seen in 12A2 and 12B2 (which are the contrast enhanced versions of A1 and B1, respectively), most features in the conventional and FCIR image closely correspond. However, when looking more in detail at for example the Roman temple (see inset), some of the walls are more distinct in the FCIR, particularly the central division. Using such a recombination of channels thus often allows for a better identification of underground structures as roads, buildings and ditches. The haze penetrating capacity of NIR imaging is illustrated in Fig. 13, displaying a very high oblique visible (A) and NIR photograph (B) from a mountainous area in Central Adriatic Italy. The NIR wavelengths clearly suffer less from the atmospheric scattering, resulting in an image with a higher contrast and far more discernable detail in the distant parts of the scene (Fig. 13B). Besides aerial reconnaissance, other archaeological research fields benefit from this modified camera, as the subsequent examples of reflected NIR photography shortly illustrate. By comparing a normal (Fig. 14A) with a pure NIR sherd image (Fig. 14B), it can be noticed that the latter more clearly reveals the places where the clay has been exposed to increased reduction during the final phase of the firing process (1) or due to the incineration of organic material and/or decomposition of calcite inclusions (2). 3098 G. Verhoeven / Journal of Archaeological Science 35 (2008) 3087–3100 Fig. 16. Difference between the visible (A) and NIR rendering (B) of the inventory number on a decorated wall fragment of a Late Iron Age handmade pot (sherd by courtesy of Wim De Clercq – Ghent University). Figs. 15 and 16 both show how the appearance of certain, visibly attestable paint pigments can be altered in the NIR. However, whereas Fig. 15B more clearly indicates the painted portions of this Roman handmade rim fragment compared to the visible photograph (15A), Fig. 16 proves certain inks to become indiscernible if they are largely transparent to NIR or they exhibit the same reflectance properties in the NIR as the outer surface of the recipient. Although the inventory number is obvious in Fig. 16A (i.e. the visible record), the same Late Iron Age sherd seems to be unnumbered in the NIR record (Fig. 16B). The other paint traces that cover this piece of ceramic are rendered with a similar contrast in both types of imagery. Remark that, contrary to the other NIR pictures, the NIR sherds in Figs. 15 and 16 do not have a greyscale character. As illustrated in Fig. 8, all colour channels capture NIR. Working on only one of these channels always yields a greyscale result, no matter if one captures exclusively visible radiation or pure NIR wavelengths. However, taking a modified DSC’s NIR information acquired in all three Red, Green and Blue channels into account will yield a photograph with an arbitrary coloured appearance, the latter being dependent on the particular NIR response of each channel and the colour channel it is attributed to in the final image. No matter its colour, the resulting frame still is a pure NIR record. In the last example, the strength of NIR to digitally uncover obliterated writing is presented (Fig. 17). As the upper black ink – which was used to mask the underwriting – is more NIR transparent than the ink applied for writing ‘‘original’’, the latter becomes clearly visible in the digital NIR photograph (Fig. 17B). Here, the extended spectral sensitivity of the digital approach largely helps as the contrast is maximised by acquiring NIR wavelengths in the upper 950 nm range (Bearman and Spiro, 1996; Chabries et al., 2003). 6. Conclusions The combination of its error-prone and awkward workflow together with a lack of technical knowledge, a scant understanding of its potential and unfamiliarity with its principles has often left (archaeological) NIR photography in the hands of only a few experienced and specialised photographers. Although this can and will not completely change in the beginning of the 21st century, the fact that today’s DSCs are perfectly capable of NIR imaging could only be beneficial to the increasing application of this kind of archaeological photography. Using a modified, dedicated NIR DSC should deal with most of the earlier film related issues, even cutting down the cost hugely as there is no more need for tons of film, processing chemicals and printing paper. Besides, geographic coordinates can be embedded into the metadata header, while the viewing capability enables a direct feedback and the digital format eases the storage and dissemination of information. As these solid-state devices have a larger Dynamic Range (DR) and QE (Chabries et al., 2003; Har et al., 2004), they can also be used in far from optimal conditions, with more consistent results due to the omitted development stage, better exposure control and a linear response to incoming radiant intensity. Consequently, NIR photography ‘‘is at least worth trying more often than is generally done to unveil the past, for the results are fairly unpredictable’’ (Conlon, 1973, p. 76). Acknowledgements This paper arises from the author’s ongoing Ph.D. which studies the application of remote sensing in archaeological surveys. The research is conducted with permission and financial support of the Fund for Scientific Research – Flanders (FWO) and supervised by Professor Dr. Frank Vermeulen (Department of Archaeology and Ancient History of Europe, Ghent University). Finally, the author’s colleague Karen Ryckbosch is acknowledged for proofreading the article and correcting the English where Fig. 17. Normal (A) and pure NIR (B) version of the partly obliterated word ‘‘original’’. G. Verhoeven / Journal of Archaeological Science 35 (2008) 3087–3100 3099 needed. All errors and misconceptions remain, of course, the author’s own responsibility. References ´ ´ Agache, R., 1968. Essai d’utilisation aerienne et au sol d’emulsions spectrozonales, ´ ´ ´ dites infrarouges couleurs. Bulletin de la Societe Prehistorique Française 65, 198–200. Aldrovandi, A., Bertani, D., Cetica, M., Matteini, M., Moles, A., Poggi, P., Tiano, P., 1988. Multispectral image processing of paintings. Studies in Conservation 33, 154–159. Alvrus, A., Wright, D., Merbs, C., 2001. Examination of tattoos on mummified tissue using infra-red reflectography. Journal of Archaeological Science 28, 395–400. van Asperen de Boer, J.R.J., 1966. Infrared reflectograms of panel paintings. Studies in Conservation 11, 45–46. van Asperen de Boer, J.R.J., 1969. Reflectography of paintings using an infrared vidicon television system. Studies in Conservation 14, 96–118. Baldia, C.M., Jakes, K.A., 2007. Photographic methods to detect colourants in archaeological textiles. Journal of Archaeological Science 34, 519–525. Barnes, D.F., 1958. Infrared Luminescence of Minerals. US Government Printing Office, Washington, D.C. Barnes, R.B., 1963. Thermography of the human body. Science 140, 870–877. Bayer, B.E., 20 July 1976. US Patent 3,971,068, Color Imaging Array. Beardsley, N.F., 1936. The photography of altered and faded manuscripts. The Library Journal 61, 96–99. Bearman, G.H., Spiro, S.I., 1996. Archaeological applications of advanced imaging techniques. The Biblical Archaeologist 59, 56–66. Braasch, O., 2007. Gallipoli Ahead – air survey between the Baltic and Mediterra´ ´ nean. In: Kuzma, I. (Ed.), Publikacia vznikla v ramci Centra excelentnosti SAV ´ ´ ´ ´ ´ Vyskumne centrum Najstarsıch dejın Podunajska pri Archeologickom ustave ´ SAV v Nitre. Archeologickom ustave SAV, Nitra, pp. 84–96. Bridgman, C.F., Gibson, H.L., 1963. Infrared luminescence in the photographic examination of paintings and other art objects. Studies in Conservation 8, 77–83. ¨ Bruning, A., 1939. Praktische Anwendungen der Infrarotphotographie und andere ¨ Fragen der Praxis. Archiv Fur Kriminologie 104, 19–22. Buettner-Janusch, J., 1954. Use of infrared photography in archaeological field work. American Antiquity 20, 84–87. Busch, D.D., 2007. Digital Infrared Pro Secrets. Thomson, Boston. Carter, G.A., Estep, L., 2002. General spectral characteristics of leaf reflectance responses to plant stress and their manifestation at the landscape scale. In: Muttiah, R.S. (Ed.), From Laboratory Spectroscopy to Remotely Sensed Spectra of Terrestrial Ecosystems. Kluwer Academic Publishers, Dordrecht, pp. 271–293. Chabries, D.M., Booras, S.W., Bearman, G.H., 2003. Imaging the past: recent applications of multispectral imaging technology to deciphering manuscripts. Antiquity 77, 359–372. Clegg, B., 2001. Light Years: An Exploration of Mankind’s Enduring Fascination with Light. Piatkus, London. Coffin, D., 2007. Decoding Raw Digital Photos in Linux. cybercom.net/%7Edcoffin/ dcraw/ (accessed 11.11.07). Conlon, V.M., 1973. Camera Techniques in Archaeology. John Baker, London. Cookson, M.B., 1954. Photography for Archaeologists. Max Parrish, London. Coremans, P., 1938. Les rayons infra-rouges. Leur nature. Leurs applications dans les ´ ´ musees. Bulletin des musees royaux d’art et d’histoire 10, 87–91. Curcio, J.A., Petty, C.C., 1951. The near-infrared absorption spectrum of liquid water. Journal of the Optical Society of America 41, 302–304. Daniels, A., 2007. Infrared Systems. SPIE Press, Bellingham. ¨ Deutsches Institut fur Normung, 1984. Strahlungsphysik im optischen Bereich und ¨ Lichttechnik; Benennung der Wellenlangenbereiche. DIN 5031-7:1984-01. Beuth-Verlag, Berlin. Dierks, B., 2004. Sensitivity and Image Quality of Digital Cameras. Basler, AG. www. baslerweb.com/downloads/9835/Image_Quality_of_Digital_Cameras.pdf (accessed 29.11.07). Dorrell, P.G., 1994. Photography in Archaeology and Conservation, second ed. Cambridge University Press, Cambridge. Eastman Kodak Company, 1968. Kodak chemistry. In: Smith, J.T. (Ed.), Manual of Color Aerial Photography. American Society of Photogrammetry, Falls Church, pp. 224–231. Eastman Kodak Company, 1997. Kodak Ektachrome Professional Infrared EIR Film. Eastman Kodak Company, Rochester. www.cocam.co.uk/CoCamWS/Infrared/ IRfilmdata/EIR.pdf (accessed 31.07.07). Eastman Kodak Company, 2003. Color Correction for Image Sensors. Eastman Kodak Company, Rochester. www.kodak.com/ezpres/business/ccd/global/plugins/acrobat/en/supportdocs/ColorCorrectionforImageSensors.pdf (accessed 03.12.07). Eastman Kodak Company, 2005a. Kodak KAF-8300CE image sensor. 3326 (H) x 2504 (V). Full-Frame CCD Color Image Sensor with Square Pixels for Color Cameras. Eastman Kodak Company, Rochester. www.kodak.com/ezpres/business/ccd/global/plugins/acrobat/en/datasheet/fullframe/KAF-8300LongSpec. pdf (accessed 12.03.07). Eastman Kodak Company, 2005b. Kodak Professional High-Speed Infrared Film. Eastman Kodak Company, Rochester. www.kodak.com/cluster/global/en/professional/support/techPubs/f13/f13.pdf (accessed 12/2002). ´ ´ ´ Edeine, B., 1956. Une methode pratique pour la detection aerienne des sites ´ archeologiques, en particulier par la photographie sur films en couleurs et sur ´ ´ ´ film infra-rouges. Bulletin de la Societe Prehistorique Française 53, 540–546. Foveon, 2007. Foveon. www.foveon.com (accessed 27.12.07). Fredlund, G., Sundstrom, L., 2007. Digital infra-red photography for recording painted rock art. Antiquity 81, 733–742. Freedman, R.A., Kaufman III, W.J., 2005. Universe, seventh ed. W.H. Freeman and Company, New York. Fujifilm, 2006. New FinePix S3 Pro UVIR. Get Ready to See Things . Differently. www.fujifilmusa.com/JSP/fuji/epartners/bin/Fujifilm_Finepix_S3Pro_UVIR_brochure.pdf (accessed 36.08.07). Gargano, M., Ludwig, N., Poldi, G., 2007. A new methodology for comparing IR reflectographic systems. Infrared Physics and Technology 49, 249–253. Gibson, H.L., 1962. The photography of infrared luminescence. Part I: general considerations. Medical and Biological Illustration 12, 155–166. Gibson, H.L., 1963a. The photography of infrared luminescence. Part II: investigation of infrared phosphorescence. Medical and Biological Illustration 13, 18–26. Gibson, H.L., 1963b. The photography of infrared luminescence. Part III: plotting infra-red phosphorescence decay curves. Medical and Biological Illustration 13, 89–90. Gibson, H.L., 1968. Infra-red recording. In: Engel, C.E. (Ed.), Photography for the Scientist. Academic Press, London, New York, pp. 299–361. Gibson, H.L., 1978. Photography by Infrared: Its Principles and Applications, third ed. Wiley, New York. Gumerman, G.J., Neely, J.A., 1972. An archaeological survey of the Tehuacan Valley, Mexico: a test of color infrared photography. American Antiquity 37, 520–527. Hackforth, H.L., 1960. Infrared Radiation. McGraw-Hill, New York. Hampton, J.N., 1974. An experiment in multispectral air photography for archaeological research. The Photogrammetric Record 8, 37–64. Har, D., Son, Y., Lee, S., 2004. SLR digital camera for forensic photography. In: Blouke, M.M., Sampat, N., Motta, R.J. (Eds.), Sensors and Camera Systems for Scientific, Industrial, and Digital Photography Applications, V, 19–21 January 2004, San Jose, California, USA. IS&T – SPIE, Bellingham, pp. 276–284. Hirsch, E.S., 1975. Infrared photography and archaeology: painted floors at Gournia. Archaeology 28, 260–266. Holst, G.C., 1996. CCD Arrays, Cameras and Displays. SPIE – The International Society for Optical Engineering, Washington. Image Engineering s.d. Spectral Sensitivity of Nikon D80. digitalkamera.image-engineering.de/downloads/spectral/Nikon_D80_spectral_sensitivities.pdf (accessed 14.10.07). Janesick, J.R., 2001. Scientific Charge-Coupled Devices. SPIE, Bellingham. ´ ´ Kodak-Pathe, 1979. La photographie infrarouge et ses applications. Kodak-Pathe, Koningslo-Vilvoorde. Konica s.d. KONICA Infrared 750 Black & White Film (¼Technical Data Sheet TDSB701) www.cocam.co.uk/CoCamWS/Infrared/IRfilmdata/konica750.pdf (accessed 06.03.07). Koyoma, T., 2006. Optics in digital still cameras. In: Nakamura, J. (Ed.), Image Sensors and Signal Processing for Digital Still Cameras. Taylor & Francis, Boca Raton, pp. 21–51. L.D., 1933. Infrared photography in decipherment of ancient documents. The British Journal of Photography 80, 311. ´ Lelong, C.C.D., Pinet, P.C., Poilve, H., 1998. Hyperspectral imaging and stress mapping in agriculture: a case study on wheat in Beauce (France). Remote Sensing of Environment 66, 179–191. Lyon, R.A., 1934. Infra-red radiations aid examination of paintings. Technical Studies in the Field of the Fine Arts 2, 203–212. Lyon, R.F., Hubel, P.M., 2002. Eyeing the camera: into the next century. In: The Tenth Color Imaging Conference: Color Science and Engineering Systems, Technologies, Applications, 12 November, 2002, Scottsdale, Arizona, vol. 10. IS&T – SID, Springfield, pp. 349–355. Marshack, A., 1975. Exploring the mind of ice age man. National Geographic 147, 64–89. Matthews, S.K., 1968. Photography in Archaeology and Art. John Baker, London. Milanesi, Q., 1963. Proposta di una facile metodica ausiliaria per lo studio della ceramiche di epoca preistorica e protostorica. Rivista di Scienze Preistoriche 18, 287–293. Milton, E.J., 2002. Low-cost ground-based digital infra-red photography. International Journal of Remote Sensing 23, 1001–1007. Mullikin, J.C., van Vliet, L.J., Netten, H., Boddeke, F.R., van der Feltz, G., Young, I.T., 1994. Methods for CCD camera characterization. In: Titus, H.C., Waks, A. (Eds.), Image Acquisition and Scientific Imaging Systems. SPIE, Bellingham, pp. 73–84. Murtha, P.A., 1978. Remote sensing and vegetation damage: a theory for detection and assessment. Photogrammetric Engineering and Remote Sensing 44, 1147–1158. Nakamura, J., 2006. Basics of image sensors. In: Nakamura, J. (Ed.), Image Sensors and Signal Processing for Digital Still Cameras. Taylor & Francis, Boca Raton, pp. 53–93. Nieuwenhuis, G., 1991. Lens focus shift for reflected ultraviolet and infrared photography. Journal of Biological Photography 59, 17–20. Parulski, K., Spaulding, K., 2003. Color image processing for digital cameras. In: Sharma, G. (Ed.), Digital Color Imaging Handbook. CRC, Boca Raton, pp. 728–757. Powlesland, D., Lyall, J., Donoghue, D.M., 1997. Enhancing the record through remote sensing: the application and integration of multi-sensor, non-invasive remote sensing techniques for the enhancement of the sites and monuments record. Heslerton Parish Project, N. Yorkshire, England. Internet Archaeology 2. intarch.ac. uk/journal/issue2/pld_index.html (accessed 21.01.08). Rawling, S.O., 1946. Infra-Red Photography, third ed. Blackie & Son Limited, London, Glasgow. 3100 G. Verhoeven / Journal of Archaeological Science 35 (2008) 3087–3100 Sony Corporation s.d. ICX205AL, Diagonal 8mm (Type 1/2) Progressive Scan CCD Image Sensor with Square Pixel for B/W Cameras, Sony Corporation. www.theimagingsource.com/products/cameras/sensors/icx205al.pdf (accessed 03.12.07). Strandberg, C.H., 1967. Photoarchaeology. Photogrammetric Engineering 33, 1152–1157. Theuwissen, A.J.P., 1995. Solid-State Imaging with Charge-Coupled Devices. Kluwer Academic Publishers, Dordrecht. Toyoda, K., 2006. Digital still cameras at a glance. In: Nakamura, J. (Ed.), Image Sensors and Signal Processing for Digital Still Cameras. Taylor & Francis, Boca Raton, pp. 1–19. Verhoeven, G.J.J., 2007. Becoming a NIR-sensitive aerial archaeologist. In: Neale, C., Owe, M., D’Urso, G. (Eds.), Remote Sensing for Agriculture, Ecosystems, and Hydrology IX, Florence, Italy, 17–19 September 2007. SPIE, Bellingham, pp. 333–345. Verhoeven, G.J.J. It’s all about the format – a raw picture of photographing RAW. International Journal of Remote Sensing, submitted for publication. Verhoeven, G.J.J., De Smet, P., Poelman, D., Vermeulen, F. Spectral characterisation of a digital Camera’s NIR-modification to enhance archaeological revelation. Remote Sensing of Environment, submitted for publication. Waldman, G., 2002. Introduction to Light: The Physics of Light, Vision, and Color. Dover Publications, New York. Wiegand, C.L., Gausman, H.W., Allen, W.A., 1972. Physiological factors and optical parameters as a basis of vegetation discrimination and stress analysis. In: Proceedings of Operational Remote Sensing, 1–4 February 1972, Houston, Texas. American Society of Photogrammetry, Falls Church, pp. 82–102. van de Wiele, F., 1976. Photodiode quantum efficiency. In: Jespers, P., Van de Wiele, F., White, M. (Eds.), Solid State Imaging: Proceedings of the NATO Advanced Study Institute on Solid State Imaging, 3–12 September 1975, Louvain-la-Neuve, Belgium. Noordhoff, Leyden, pp. 47–90. Young, H.D., Freedman, R.A., 2004. Sear and Zemansky’s University Physics: with Modern Physics, 11th ed. Pearson, San Francisco. Ray, S.F., 1999. Scientific Photography and Applied Imaging. Focal Press, Oxford. Ray, S.F., 2002. Applied Photographic Optics: Lenses and Optical Systems for Photography, Film, Video, Electronic and Digital Imaging, third ed. Focal Press, Oxford. ¨ Reichstein, J., 1974. Schwarz-Weiß-Infrarotphotographie als Hilfsmittel fur die Analyse schwer beobachtbarer Befunde. Offa 31, 108–125. Richards, A., 2001. Alien Vision: Exploring the Electromagnetic Spectrum with Imaging Technology. SPIE, Bellingham. Rieke, G.H., Visnovsky, K., Swarthout, K., 1994. Detection of Light: From the Ultraviolet to the Submillimeter. Cambridge University Press, Cambridge. ´ ´ ´ ´ Rigaud, P., Bouyer, P., 1986. Archeologie aerienne: teledection infrarouge par ` photographies verticales a partir d’un ballon captif et d’un avion. Travaux ´ d’archeologie limousine 7, 7–12. Rørslett, B., 2004. All You Ever Wanted to Know About Digital UV and IR Photography, But Could Not Afford to Ask. www.naturfotograf.com/UV_IR_rev00.html (accessed 10.09.06). Rørslett, B., 2007. Lens Survey and Subjective Evaluations. www.naturfotograf.com/ lens_surv.html (accessed 11.12.07). Ross, K., 1933. Zur Photographie mit ultraviolettem und infrarotem Licht. Zeiss Nachrichten 4, 19–27. Rutherford, A., 1977. Infrared photography using an ultraviolet radiation source. Medical and Biological Illustration 27, 35–36. Schiffman, L., 1995. Reclaiming the Dead Sea Scrolls. Doubleday, New York. Silberberg, M.S., 2006. Chemistry: The Molecular Nature of Matter and Change, fourth ed. McGraw-Hill, Boston. Simmons, H.C., 1969. Archaeological Photography. University of London Press, London. Slater, J.C., Frank, N.H., 1974. Electromagnetism. Dover Publications, New York. Smith, G.S., Zimmerman, M.R., 1975. Tattooing found on a 1600 year old frozen, mummified body from St. Lawrence Island, Alaska. American Antiquity 40, 433–437.