Scanners - a survey of current technology and future needs

Transcription

1 Research Collection Conference Paper Scanners - a survey of current technology and future needs Author(s): Baltsavias, Emmanuel P.; Bill, Ralf Publication Date: 1994 Permanent Link: Rights / License: In Copyright - Non-Commercial Use Permitted This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use. ETH Library

2 SCANNERS - A SURVEY OF CURRENT TECHNOLOGY AND FUTURE NEEDS Emmanuel P. Baltsavias*, Ralf Bill** *Institute of Geodesy and Photogrammetry, Swiss Federal Institute of Technology ETH-Hoenggerberg, CH-8093 Zurich, Switzerland Tel.: , Fax: , manos@p.igp.ethz.ch **Institute for Geodesy and Geoinformatics Department of Land Cultivation and Environmental Protection, University of Rostock Justus-von-Liebig-Weg 6, D Rostock, Germany Tel.: , Fax: ISPRS Commission I, Working Group 5 KEY WORDS: Image Scanners, Scanner Technology, Photogrammetric Scanner Requirements, Scanner Specifications ABSTRACT Scanners are necessary for retrieving digital information from analogue imagery or maps. In the context of this paper, we will discuss scanners as input devices for softcopy photogrammetric systems, which ask for high radiometric and geometric quality for the scanning process. Thus, the major emphasis in this paper lies on photogrammetric scanners. In addition, some cheaper alternatives are also discussed. This paper presents a classification and overview of scanners, some important scanner aspects and requirements, and scanner specifications for photogrammetric tasks. 1. INTRODUCTION Scanners are an essential part of softcopy photogrammetric systems. Although the developments in direct digital data acquisition have been enormous in the last decade, film-based systems are used in all fields of photogrammetry. In aerial photogrammetry film-based systems will provide the main data input for many years to come. Filmbased satellite images are provided by many Russian sensors, while also in USA companies, previously active in military applications, want to launch film-based high resolution commercial systems. The main use of scanners today is definitely in the digitisation of aerial images. The main applications that increase the need for digital aerial data are (i) the digital ortho-image generation, (ii) the automatic DTM generation, (iii) generation and update of digital feature databases and (iv) the integration of digital data, particularly DTMs, ortho-images and derived products, in GIS. Scanners of documents (reflective and/or transparent) can be classified according to their function in the following categories: 1. Photogrammetric scanners 2. Modified analytical plotters or monocomparators 3. Scanners of large documents 4. Microdensitometers 5. DeskTop Publishing (DTP) scanners 6. Other scanners such as scanners of documents and 3-D objects, slide scanners, text document scanners, multiple purpose scanners (e.g. scanner/copier/colour printer, scan/edit/fax scanners), specialised scanners (e.g. hand-held scanners, engineering document scanners, roentgen-image scanners, microfiche digitisers, barcode scanners) The paper mainly concentrates on photogrammetric scanners. Alternatives, especially DTP scanners, are also investigated. Assuming that for photogrammetric applications, a scanner should be able to scan aerial images (23 x 23 cm) with a minimum optical resolution of 600 dpi and sufficient radiometric and geometric accuracy, only scanners of the first five categories should be addressed.

3 1.1. Scanner Classification Scanners can be classified according to different criteria: A. Dimensionality of simultaneously sensed elements point sensors (one pixel scanned at a time) They are often used in microdensitometers and drum scanners for high geometric and radiometric resolution. Generally each pixel is illuminated by a laser beam and the sensor consists of a photomultiplier tube (PMT). PMTs have high speed and high dynamic range. Microdensitometers and some drum scanners have a great flexibility in selecting different pixel sizes and forms and x-, y-spacing between pixels. Some DTP drum scanners use cheaper illumination/sensor systems, e.g. halogen lamps and usually photodiodes (their dynamic range is less than that of PMTs but higher and with a more linear response than that of CCDs). line sensors The line usually consists of CCD elements but also photodiodes or charge-coupled photodiodes. There might be one linear CCD, or more, whereby in the latter case the linear CCDs are optically butted using beam splitting prisms or rotating mirrors. For colour scanning one or multiple 3-chip linear CCDs may be used. area sensors They consist of CCD chips with a resolution ranging from 512 x 512 to 3000 x 2300 pixels. B. Scanning pattern/movement drum scanners Generally both stage and sensor are moving. flatbed scanners The following cases can be distinguished: moving stage/stationary sensor, stationary stage/moving sensor, both sensor and stage stationary. To achieve high geometric resolution either the stage or the sensor is moving, whereby the movement can be in two or one direction. Movement in two directions can be realised by all type of sensors (point, line, area), movement in one direction only by linear CCDs with many sensor elements. The disadvantage of the first case is that it requires high geometric accuracy in two directions. In addition, in the case of line and area sensors there are often clearly visible radiometric differences along the seam lines of neighbouring line stripes or area patches due to illumination instabilities and different sensor element response. The a posteriori correction of this problem requires scanning of neighbouring lines/ patches with overlap and application of a radiometric equalisation procedure as in mosaicking. The documents are usually scanned by a mechanical movement, however, electronic deflection of the beam or optical scanning (oscillating mirrors, rotating prisms) can also be used. The light source either illuminates the whole object to be scanned or only the portion that is scanned each time (thus, if the sensor is moving the light source must also move synchronously) Overview of market and different scanner types Photogrammetric scanners are produced by companies involved in photogrammetry, are flatbed scanners employing linear or area CCDs, have a high geometric accuracy (typically 2-5 µm), high geometric resolution ( µm minimum pixel size), and sometimes software for interior orientation and scanning in the photo coordinate system. Generally they are tightly coupled to digital photogrammetric workstations running different applications software, and some of them support film roll processing. They can be divided into two groups based on price: these priced higher than 125,000 $ (Zeiss/Intergraph PS 1, Leica/Helava DSW 100 and DSW 200), and these in a price range of 40,000-75,000 $ (ISM s DiSC, Vexcel VX 3000, Wehrli s RM 1). A overview of photogrammetric scanners currently in the market is given in Table 1. Scanning by using analytical plotters equipped with CCD cameras never became popular, and it is very unlikely that this will change. Analytical plotters are widespread, have a high geometric accuracy, and are equipped with photogrammetric software. However, installation of CCDs at analytical plotters require modifications that are very often expensive, the cost of the hardware (cameras, framegrabbers, monitors) and the necessary calibration software is not inexpensive either, the scanning process is slow, and the radiometric quality suffers because illumination, optics, filters etc. are not optimised for scanning. Additionally, the concept of a scanner does not fit very well to analytical plotters which, by their very nature, are operator driven, and apart from that scanning would obstruct the normal work at the analytical plotter. Scanners of large documents are in most cases drum scanners, and thus their geometric accuracy is generally less than that of flatbed scanners. They may have a high radiometric resolution, but often they do not scan transparencies. Often they can also plot, and thus are quite expensive. Their main application is in scanning of large format plans and maps, and less in scanning of films. Conventional microdensitometers (Perkin-Elmer, Joyce- Loebl) have a very high geometric resolution (up to 2 µm), a very good geometric accuracy and radiometric quality (12-16 bit) but they are not very widespread, are generally very expensive, slow, quite complicated to calibrate, and less flexible both hardware- and software-wise. They are important for analysis of micro-image properties and specific applications where radiometric resolution and accuracy are of major concern (astronomy, medicine, image quality determination).

4 Table 1: Photogrammetric scanners Brand Zeiss/Intergraph Leica/Helava Leica/Helava Int l Systemap Corp. Vexcel Imaging Corp. Wehrli and Assoc. Inc. Model PhotoScan 1 DSW 200 DSW 100 DiSC VX 3000 RM-1 Mechanical movement flatbed, flatbed, moving stage * stationary stage flatbed, stationary stage flatbed, stationary stage vertical back-lit stage, moving sensor/optics flatbed, moving stage Sensor type linear CCD, 2048 pels Kodak Megaplus 2029 x 2044 CCD 2 standard area CCDs (high, low resolution) 3-chip colour CCD, 8,000 pels standard area CCD linear CCD, 2048 pels Scanning format (mm) 260 x x x 254 (270 x 270 option) 320 x x x 260 Geometric resolution (µm) (and any multiples of 2) 10 (high res.) (low res.) Radiometric resolution (bits) (internal/output) 10 /8 10/ / Illumination halogen, 100W, fiber optic xenon, fiber optic, sphere diffusor quartz halogen, fiber optic, sphere diffusor halogen, fiber optic cold cathode, variable intensity fluorescent Colour passes Density range 3D D D Geometric accuracy (µm) 2 < Radiometric accuracy (DN) ± 2 ± 2 Scanning throughput and/or speed variable, 2 Mb/s (7.5 µm) 0.5 Mb/s (15 µm) 1 Mb/s (12.5 µm) (incl. disk save) max 35 mm/s 67 Kb/s max 35 mm/s Mpixel/s variable, 31 Kb/s ** (incl. set-up and disk save) 0.2 Mb/s Host computer/ Interface Intergraph UNIX workstation Sun Sparc 20 Unix-based PC/ SCSI PC-DOS/ SCSI-2 Unix-based PC and workstation required PC-DOS/ SCSI Price (SFr.) 400,000 ca. 240,000 ca. 240,000 ca. 115,000 ca. 90,000 ca. 80,000 * Stage refers to the stage holding the document Special option to scan 30 cm x 30 cm Russian satellite imagery Typical values. Highest resolution is 7 µm. Through a software option any pixel size can be selected ** About twice this speed can be achieved with the model

5 Table 2: DTP scanners Brand Agfa Sharp Scitex Intergraph Model Horizon Plus JX-610 Smart 340 L ANA Tech Eagle 1760 Mechanical movement flatbed, stationary stage flatbed, moving stage flatbed flatbed Sensor type 3 linear CCDs, 3 x 5,000 pels linear CCD, 7500 pels linear CCD 2 linear CCDs, 2 x 5,000 pels Scanning format (mm) A3 (reflective) 240 x 340 (transp.) 305 x 432 A3 419 x 610 Geometric resolution (µm) (v) x 42.3 (h) * Radiometric resolution (bits) (internal/output) 12/10 or 8 12/8 8 8 Illumination halogen, 400 W 3 RGB strobing fluorescent lamps quartz halogen, fiber optic Colour passes 3 1 Geometric accuracy (µm) 80 (without calibration) 460 (in x) 0.1% (in y) Scanning throughput and/or speed 0.35 Mb/s (1200 dpi) * mm/s 0.62 Mb/sec (A3, 600 dpi) 0.48 Mb/s (A4) 0.68 Mb/s (A3) Host computer/ interface Mac, PC, Unix workstations /SCSI-2 Mac, PC, Unix workstations /GPIB, SCSI-2 Mac PC. PS-2, Mac Price (SFr.) 45,000 22,000 48,000 * Horizontal is in CCD direction, vertical in scanning direction Values estimated by the first author for Agfa Horizon (32 Mb internal image buffer) connected to a Sparc 2 DTP scanners have been developed for applications totally different than the photogrammetric ones. However, since they constitute the largest sector in the scanner market, they are subject to rapid developments and improvements. The consultancy BIS Strategic Decisions (Norwell, MA) forecasts that the colour flatbed DTP scanner market will grow 39 % annually over the next five years (Holch, 1993). While DTP scanners include both flatbed and drum ones, only flatbed scanners will be considered here. Although drum scanners have a high geometric resolution (up to 4000 dpi), and high density range (up to about 4D), they are more expensive than their flatbed counterparts (starting at ca. 40,000 $ for formats larger than A4), and most importantly they have low geometric accuracy due to drum inaccuracies, unflatness of film on drum etc. and because of the same problems an accurate geometric calibration is not feasible. Flatbed scanners typically employ one or more linear CCDs, and move in direction vertical to the CCD to scan a document. They can scan binary, grey level and colour data (with one or three passes), may have good and cheap software for setting the scanner parameters, image processing and editing, and can be connected to many computer platforms (mainly Macs and PCs, but also Unix workstations) via standard interfaces. They can usually scan A4 format, but some can scan up to A3 or even more. Some do not scan transparencies, others do so but only of smaller format. There exist a handful of scanners which can scan aerial imagery films (characteristic representatives are the 1200 dpi Agfa Horizon Plus and Horizon, and the 600 x 1200 dpi Sharp JX 610 and 600). For more details on DTP scanners see Holch, Flatbed scanners have a resolution of up to 1200 dpi (21 µm pixel size) over the whole scan width. Few scanners offer the option to increase the resolution (e.g. up to 2400 dpi) by projecting a document portion (smaller than the full width) on the CCD. Their price range, with few exceptions, is 1,000-25,000 $. The big price jump occurs when going from A4 to A3 format. The transition from 600 dpi to 1200 dpi

6 costs less. A3 scanners with 600 x 1200 dpi start at ca. 12,000 $. A4 scanners with 600 x 1200 dpi and transparency options cost much less (e.g. the Ricoh FS 2 with 190 x 297 mm transparency option costs ca. 3,500 $). Table 2 shows the major features of DTP scanners that fulfil the previously mentioned scanner requirements for photogrammetric applications. Their radiometric resolution and quality, and scanning speed can be comparable to or even exceed that of the more expensive photogrammetric scanners. DTP scanners with automatic density control and user definable tone curves that can be applied during scanning need for the setting of the scan parameters a few minutes as compared to more time (even one hour) required by some photogrammetric scanners. In particular, the sensor chip and the electronics of DTP scanners are updated faster and are in most cases more modern that the respective parts of photogrammetric scanners. The main disadvantage of DTP scanners is the insufficient geometric accuracy and stability, caused mainly by mechanical positioning errors and instabilities, large lens distortions, and lack of geometric calibration software Literature overview It is surprising how little has been reported up to now about the performance of scanners. As an example, although the Zeiss/Intergraph PhotoScan 1 was introduced in 1989 and Helava s HAI 100 (the predecessor of Leica/ Helava DSW 100 and DSW 200) even earlier, there are very few published independent reports on the performance of these scanners. This is even more surprising since the cost of such scanners is high, while for other much cheaper input devices, like CCD cameras, there are hundreds if not thousands of published papers. In the following an overview of publications on scanners and related subjects will be given. A survey of scanners (but with incomplete data and omitting many scanners) is published in Geodetic Information magazine, January Aslund et al., 1977 describe the IRIS comparator/scanner, and Hanf and Deter, 1983 the FEAG scanner/plotter, Brown, 1987 reports on the Autoset scanner/comparator for industrial photogrammetry, Faust, 1989 on the Zeiss/Intergraph PhotoScan 1 scanner, Leberl, 1990 and Leberl et al., 1992, on the VX-series of scanners. Gerhard, 1991, describes the Signum HIRES scanner used in ortho-image production. Roos, 1993 gives the description, operation and test results of the precise image scanner IDS (Image Digitising System, a scanner resembling the Zeiss/Intergraph PS 1) which was installed at the US Army TEC. A description of various scanners, practical experiments with them and recommendations for testing procedures are included in the proceedings of the OEEPE-ISPRS Joint Workshop on the Analysis of Photo- Scanners, February , in Lausanne (the proceedings are available from Prof. O. Koelbl, EPFL, Laboratory of Photogrammetry, GR-Ecublens, CH-1015 Lausanne, Switzerland). Colomer, 1993, Dueren et al., 1993, and Kiefer, 1993 report on the integration of scanners in mapping organisations and present some of their characteristics and problems. Accuracy analysis of scanners are published by Johannsen, 1976, and Lichtner, Gruen and Slater, 1983 present a strategy for testing the radiometric and geometric performance of high resolution scanners. Boochs, 1984 reports on tests and accuracy analysis of drum scanners mentioning that they exhibit significant geometric errors. Bethel, 1993, 1994 presents a short description and calibration results for the Wehrli RM 1 scanner. Miller and Dam, 1994 present standards for image scanners used in digital photogrammetry and give some calibration results for the Leica/Helava DSW 200. Thompson and Quellette, 1972 discuss the accuracy requirements of scanners. Makarovic and Tempfli, 1979 report on image digitisation for automatic photogrammetric processing. Ehlers, 1991 reports on digitising and photogrammetric requirements. Diehl, 1992, discusses the optimal digitisation steps for usual films. Boberg, 1992 presents results from a subjective evaluation of image quality, and reports that image sharpness had the largest influence on quality assessment, while mean density, contrast and granularity had a smaller influence. Schroeder, 1992 discusses the spatial resolution of a linear CCD in terms of the optical transfer theory. Baehr, 1988 presents different methods for measuring the geometric resolution of digital cameras in general and in relation to specific applications. A very good, theoretical book on different aspects of imaging, including a chapter on microdensitometry, is written by Dainty and Shaw, An easy-to-read but not theoretical and detailed book dealing with electronic image acquisition and scanning technology is written by Larish, There are hundreds of papers on geometric and radiometric analysis, and error modelling of CCDs and many of them are useful, as most scanners employ linear or area CCDs. Because of space limitations we will restrict ourselves to Beyer, 1992 where many radiometric and geometric tests are presented, as well as a long reference list. A comparison of flatbed DTP scanners is given by Matazzoni, 1991, Diehl and Edwards, 1992, and Seiter, Bosma et al., 1989 give an evaluation of low-cost scanners. Steiger, 1991 makes a quality comparison between colour DeskTop Reproduction and Electronic Image Processing scanners. Sarjakoski, 1992 reports on the Sharp JX 600 DTP scanner and the development of a calibration procedure with an accuracy of 0.2 pixel (8 µm) - he also reports in other tests an accuracy of 0.1 pixel. His results are too optimistic because all blunders larger than 30 µm had been removed (actually often errors of +/- 60 µm occurred). Klaver and Walker, 1992 report that the same scanner without calibration has sometimes local errors up to 170 µm. Chen and Schenk, 1992 present an extension of the DLT for calibration of linear and area CCDs, and scanners. Huurneman, 1992 presents a low-

7 cost transparency scanner made from off-the-self components, its calibration and accuracy. A description and testing of the Agfa Horizon is given in Baltsavias, Investigations on low-cost peripheral devices for digital photogrammetric systems are presented by Cramer et al., Guelch, 1986, Baltsavias, 1988, Wilkins 1990, Fuchs and Ruwiedel, 1992, deal with CCDs mounted at analytical plotters, their use as scanners, and their calibration. 2. SCANNER ASPECTS AND REQUIREMENTS FOR PHOTOGRAMMETRIC APPLICATIONS Different scanner aspects and necessary requirements for standard photogrammetric tasks will be discussed below. They are important for both users and scanner vendors. Knowledge on these topics allows users to better understand and evaluate scanners or appropriately set the scanning paparemeters. Vendors can use this information in the stage of design and construction of a scanner or the update of an existing model. Different implementation options and technological alternatives will be presented Illumination The illumination must be high in order to achieve a better radiometric quality and higher SNR. This is due to the high scanning speed and the light intensity loss in the parts of the optical path. The higher the scanning speed, the higher the illumination should be since the dwelling time (integration time for CCD sensors) is reduced. Different parts of the optical path (filters, beam splitter) lead to intensity losses (for the Agfa Horizon with 400 W halogen lamps only the equivalent of 100 mw light reaches the CCD surface). Particularly in blue the power of the illumination is dramatically reduced. On the other hand, high power light sources generate heat, which must be treated appropriately, in order to minimise the influence on the mechanical parts and the electronics (cooling, use of cold light, placement of the light source away from the sensitive scanner parts and use of fiber optics for light transfer). The spectral properties of the light source and its temporal stability (related also to the power supply stability) are important factors. In some scanners the light source has variable intensity in order to obtain balanced colour scanning (highest intensity used for blue channel, lowest for red). The illumination should be uniform over the whole field of view of the sensor and preferably diffuse (not directed). Diffuse illumination can be accomplished by use of fluorescent lamps, diffuser plates in front of the light source, diffuse reflectors, and Ulbricht spheres. Light sources usually include halogen and fluorescent lamps (often over 100 W), as well as laser beams Dynamic range and quantisation bits The dynamic range of films can be in extreme cases very high (e.g. 16,000:1). To capture this information a quantisation up to 16-bit would be necessary. For aerial images 10 to 12 bit quantisation suffices to capture the information even in difficult scenes containing very bright and dark regions. This dynamic range can be supplied by point sensors and linear CCDs, and by special purpose area CCDs which are however not used in scanners (an exception is the Leica/Helava DSW 200 which uses the better quality Kodak Megaplus KAF-4200 sensor). Apart from microdensitometers, there are photogrammetric, drum and DTP scanners that have A/D converters with bit quantisation, but since almost all software and hardware supports only 8-bit/pixel and to avoid problems with excessive amount of data and image display, the data is reduced to 8-bit. The user often has no influence and no information on how this reduction is made. If this is done properly, then the result will be a radiometrically better image with higher SNR. However, 10 or 12 bit quantisation can lead to an improvement only for low noise levels. This is not always the case, particularly with CCD based scanners. Parameters like control of heat generation to reduce thermal noise, maximum charge storage capacity, integration time, smearing etc. are not always optimised to allow a truly beneficial 10 or 12-bit quantisation. As an example consider a standard area CCD with a maximum storage capacity of 100,000 electrons per sensor element, and a typical noise level of several hundred electrons. Then, if a linear quantisation of the grey levels is applied using 12- or 10-bit quantisation, a grey level would correspond to 24 or 98 electrons respectively. In both cases, such a fine quantisation does not make sense since the noise level is much higher than one grey level. Thus, the 10/12-bits are sometimes used as a selling argument but often they do not reflect an essential quality difference to 8-bit scanning. The scanner should be able to accommodate densities in the range D. Density close to 0 is necessary when scanning glass plates (films start at D), while a density over 3D is required for B/W images with very high dynamic range and many colour images Colour scanning Colour scanning can be implemented by: primary or complementary colour filters spatially multiplexed on the sensor elements (1-chip colour linear or area CCD) use of 3-chip CCDs (linear or area arrays) rapidly strobing fluorescent lamp and dichroic filters, halogen ray and rapidly rotating filter wheel, flashing 3-colour fluorescent lamps use of filters before the sensor (RGB and neutral filters) or rotating lamps

8 The first three approaches require one scan, while the last one three. The first approach leads to reduced spatial resolution and sometimes pattern noise in the image, and it lacks the ability to colour balance (blue in particular). The second is the best approach but also most expensive. Although many claim that the third approach is faster than the fourth, the scanning time is similar, if the same integration time for each colour is required. Another general belief that, the third approach often leads to smearing, while the fourth might suffer from misregistration of the three channels, is also not always correct. Misregistration between the colour channels can occur not only due to positioning errors in scanners that perform three passes but also due to the lens and other optical parts (mirrors, filters on glass plates). A problem with colour balancing will occur, if the sensitivity of the sensor is nonuniform (particularly CCDs have a lower sensitivity in the blue region as compared to green and red region of the spectrum). To avoid or reduce the problem of unbalanced colours each channel can be treated differently using one of the following approaches: variable light intensity, variable integration time, individual exposure control, individual gain factors Linear versus area CCDs Among the sensors, the most promising and widely used are linear CCDs. Today there are various linear CCDs with 5,000 to 10,000 elements (the MN3666 from Matsushita with 7,500 elements, a 3-chip colour CCD from Kodak with 2,000-8,000 elements, TCD 141 from Toshiba with 5,000 elements, the firm Technolink (Americas) sells linear chips with 10,000 elements, Thomson has linear arrays with 5184 elements, Dalsa with up to 6,000 elements, EG&G and Fairchild sell linear CCDs of various lengths, and the list is continuously growing). With current technology multiple linear CCDs can be optically butted with high precision to result in a line with sufficient elements for a high resolution scan of 10 µm or less. Although area CCDs have a larger number of elements and theoretically should lead to a faster scanning, they have several disadvantages. Large area CCDs are very expensive and have blemishes (pixels whose grey values differ a lot from the grey values of their neighbours), and standard CCDs need a lot of frames (and time) to cover the whole image and a precise positioning in two directions as compared to a long linear CCD which can scan the whole image in one swath. Radiometric differences along the seam lines of the frames can occur as explained above. Linear CCDs provide a better radiometric quality, have higher charge transfer efficiency, and suffer less from electronic noise (smearing etc.) than area CCDs. Another reason for their lower noise is the fact that they have fewer clock signals, so the latter can be better isolated from the video. Antiblooming drains which are important with high contrast objects are easier to implement with linear CCDs. They have higher dynamic range (10,000:1 is possible, compared to 1,000:1 of a very good area CCD). Area CCDs are designed to maximize the gain (the output signal) which also increases the noise, while linear CCDs are designed to maximize the dynamic range, so they have less noise and therefore respond to less input light and due to the higher dynamic range they will respond to higher light levels as well. They have adjustable integration time while area CCDs are usually locked to the RS170 or CCIR specifications (33 or 40 ms respectively). Linear CCDs have higher speed (pixel rates of up to 120 MHz) than standard area CCDs because parallel output for high data rates is easier with linear CCDs and their integration time is shorter. Each line can be accessed immediately after integration, while with area CCDs all preceding lines must be first read out, and they are usually interlaced (so the first even line can be output only after all odd lines have been read out). Since a line contains a single row of pixels, the uniformity can be held tighter than in an area array with several hundred-thousand pixels. They also have a more precise geometric centring. In very high precision imaging applications, contrast correction hardware and software algorithms are more easily implemented over a single line of pixels. For some of the advantages of linear CCDs see EG&G, 1993 p. 359 and 389. Linear CCDs also have some disadvantages. Normal operation of linear CCDs results in much shorter integration times than that of area CCDs (typically 1 ms), and therefore a much higher light intensity is required. Linear CCDs due to their long length place special demands upon lenses and associated optics. They usually have smaller pixel size than the area CCDs, thus usually smaller maximum charge storage capacity. When applying subsampling, linear CCDs lead to a slight image degradation of horizontal lines (parallel to the CCD) as compared to vertical ones due to image smear caused by the high scanning speed Scanning speed Many users are fascinated by high speed, and vendors of high speed scanners use this as a selling argument. First of all, the total time for a successful scan should be taken into a account. As an example, the Agfa Horizon needs ca. 2 min (for 1200 dpi) to scan, transfer to swap disk space of the host computer and display in a window, a 30 Mb image (Horizon with a 32 Mb image buffer connected to a Sun Sparc 2). Theoretically, the scanner can mechanically move the 15,000 pixels long CCD line at up to 100 mm/s, which means for a 1200 dpi resolution a grey level image of 15,000 x 4724 pixels could be scanned in 1 sec., i.e. a 71 MHz bandwidth. The electronics of the scanner have however a maximum bandwidth of 15 MHz, so the maximum scanning speed that can be realised for 1200 dpi is 21 mm/s. The same realisable maximum scanning speed is enforced because of the signal integration time of 1 ms, i.e. in one second 1000 pixels with µm pixel size can be scanned. Actually Horizon scans with 1200 dpi at ca mm/s, i.e. a 9 MHz bandwidth. With this bandwidth the 30 Mb image can be scanned in 3.3 sec. The

9 majority of the required 2 min is for transfer of the data via the SCSI interface to the host and for saving the data on disk. That means that the physical scanning process could be much slower without increasing considerably the overall time, i.e. with a 18 times slower scanning rate (0.5 MHz, speed of 0.7 mm/s) the overall time would be 3 instead of 2 min. This is still not the total time needed for a successful scan. The time needed to set and optimise the scanning parameters can be more than the time required to do the final scan. In certain cases, as with the Agfa Horizon, a successfully scanned image must be transferred from the swap disk space to the disk space allocated to the users. For the Agfa Horizon this procedure needs 5 min for a 30 Mb image, so it is 2.5 times more than the time required for a successful scan. The digitisation of the image is just one part in the processing chain. Usually other processes follow, like ortho-image and DTM generation, mapping etc., i.e. procedures that require more time than the scanning itself. As a conclusion, the scanning speed could be slower without any significant reduction in production throughput. The reduction of the scanning speed would have several advantages: the scanning mechanism (mechanical, optical, etc.) could be slower which means simpler, cheaper and stabler components ; the integration time could be increased which means higher signal to noise ratio, and no need for powerful illumination which is expensive and generates a lot of heat, influencing the optomechanical and electronic parts, and requiring mechanisms for controlling the heat dissipation ; vibrations in the scanning direction could be avoided or reduced ; the smear in the moving direction depends on the product (scanning speed x integration time), so it could be decreased if the integration time is increased less than the scanning speed is decreased; noise like lag which is typical of high speed imagers could be decreased ; the bandwidth and the price of the electronics could be decreased while more operations could be applied in real-time using hardware processing capabilities ; large image buffers in the scanner that are sometimes required to store the data before transferring it to the host would not be necessary since the low data rate could be accommodated by the host/scanner interface or a small image buffer Optimal geometric resolution The question of optimal geometric resolution is open and much, partly contradicting, has been published on the topic. Aerial films contain a large amount of valuable information for many different purposes. It is true that to preserve the resolution of the original analogue film an appropriate pixel size should be used. For high resolution film and aerial cameras with forward motion compensation a resolution of 60 lp/mm can be achieved (Meier 1984) and this would require a pixel size of about 6 µm (assuming 2.8 pixels are necessary to resolve a line pair), resulting to a 1493 Mb digital file. Even state-of-the-art high performance workstations have difficulties in handling this amount of data. Therefore, in practice certain compromises are made by the users. It has often been reported that users (incl. mapping organisations, big private companies and universities) despite having scanners with higher than 10 µm resolution, actually scan with a pixel size 2 to 4 times larger (Colomer, 1993, Kiefer, 1993). The optimal geometric resolution clearly depends on the application and the user requirements. For DTM generation, there are many published good results that were achieved with pixel sizes of µm. Krzystek and Wild, 1992 report accuracy comparisons for DTM generation between images scanned with 15 and 30 µm at a Zeiss/Intergraph PhotoScan 1. The accuracy difference between the two versions was small and the 30 µm version had an accuracy of ca. 0.1 of the flying height over ground. For measurement of fiducials which can be measured with least squares matching with an accuracy of less than 0.1 pixel, a resolution of 1200 dpi completely suffices, 600 dpi is just adequate. Problems may occur with signalised control points. With their standard size (optimised for the measuring mark of an analytical plotter) a 1200 dpi resolution is almost the minimum for their accurate measurement (for small image scales and small targets a resolution higher than 1200 dpi is required). In the recent OEEPE test on digital aerotriangulation (image scale 1:4,000) a sigma (after bundle adjustment) of ca. 5 and 6.5 µm was achieved for 15 and 30 µm scan pixel size respectively. In many ortho-image projects aerial imagery scanned with a pixel size of 25 µm or more has been used. For ortho-image generation many people claim that the resolution can be fairly low, e.g µm or even lower. The argument they use is that for creating a sufficient quality hardcopy with accuracy similar to that of printed maps or even lower this resolution suffices (Leberl, 1992). However, the digital ortho-images can be used, and there lies their value, in applications that require higher resolution and accuracy, as in updating of topographic maps. For high geometric accuracy, a dpi resolution may be sufficient, and 1200 dpi allows detection of most cultural features (even for a small scale of 1:40,000 features of ca. 1 m can be detected) but for visual interpretation an even higher resolution might be desirable especially in urban and forested areas (Skalet et al., 1992). However, the visual interpretation can also be improved by other means than resolution, like digital image enhancement, use of colour, oversampling, and stereo viewing. In addition, the overall image quality does not only depend on pixel size. The quality depends on the characteristics of the whole imaging system including illumination, optics, filters, sensor, A/D converter, hardware and software processing, quality of the scanned document etc. and to overestimate the importance of pixel size would be wrong. Additionally, a very small pixel size leads to an increase of radiometric noise because of many reasons, among which an important one is the granularity of the film. Le Poole, 1992 claims that the MTF of films is rather the same for different emulsions (grain sizes) with a value of around 50 % or less at 40 lp/mm. Thus, he concludes that for a faith-

10 ful digitisation a 10 µm pixel size is needed. Leberl et al., 1992 present different arguments regarding the optimal pixel size and suggest for 30 lp/mm imagery a pixel size of µm (based on the assumption that at most 3.3 pixels are needed to resolve a line pair). They conclude that if all photogrammetric applications should be accommodated by the scanned imagery, then a pixel size of up to 8.5 µm is needed, although arguments can be made that µm are sufficient. Leberl, 1992 presents a slightly different opinion using the assumption that 2.8 pixels are needed to resolve one line pair (thus, for 30 lp/mm a pixel size of 12 µm would be needed). Baehr, 1992 discusses the appropriate pixel size for ortho-images and proposes the use of variable pixel size within the same image and depending on the application. Diehl, 1990, discusses the effect of graininess on the radiometric noise in digitised images. He states that for 7.5 µm pixel sizes the radiometric noise due to graininess can amount to more than 20 % of the signal, while for larger pixel sizes it is much less. Hempenius and Xuan, 1986 report on optimal parameters for digitising aerial images. They claim that less than 2.8 pixels are necessary to resolve a line pair and suggest a 25 µm pixel size, whereby in their discussion they consider only the interpretation aspects and not geometric positioning accuracy. Trinder, 1987 mentions that a pixel size of 25 µm is adequate for pointing and interpretation, but systematic measuring errors of 0.2 pixels (however with manual pointing) require a pixel size of 10 µm in order to preserve the precision of the image geometry. Trinder, 1989 uses synthetic images and reports tests on the influence of image quality, quantisation level, target size in pixels, and SNR on the precision of target location. Systematic errors can be significant if there is substantial asymmetry in the target intensity profile. Mikhail, 1992 reports on tests of the geometric accuracy of different photogrammetric operations using digitised images and the Leica/Helava DPW and compares the results to the accuracy achieved with a DSR analytical plotter. Images scanned with 25 µm gave similar results to the results of the DSR, for certain applications even a 50 µm scan pixel size was sufficient. Summarising, there is no clear answer to the question of optimal scanning geometric resolution. The scanning vendors tend to argue that the optimal resolution is very close or identical to the highest resolution of the scanners they are selling. Among the scientists and users there is also no agreement. Decisive factors as to what is the optimal resolution are the applications of the user, and the amount of data that can be handled. Although there are rapid developments in computer technology, large data sets resulting from high scanning resolution can still not be handled conveniently or not at all. Today the limit for a practical handling and interactive work seems to be around µm. Addressing the topic of the optimal geometric resolution from a practical and realistic point of view, it seems that a resolution of 1200 dpi can suffice for all photogrammetric tasks. The main limitation is the interpretation of fine details, and the measurement of signalised control points especially for high altitude flights, but by using large natural control points whose 3-D coordinates can be determined with GPS or measured in existing maps this problem can be circumvented Zoom-in capabilities It is definitely desirable to have a scanner that can scan at different resolutions. This can refer (i) to different resolutions for different images, and (ii) different resolutions within one and the same image. The first case is simpler and practically offered by almost all scanners. The different resolution can be offered by the following means: 1. Optical zoom This also requires refocussing each time the resolution changes. It can be implemented either with very stable and precise optomechanical systems or by systems that can be self-calibrated, e.g. by means of a réseau. Two cases can be distinguished: (i) the resolution increases by a respective decrease of the scan dimensions, i.e. the same number of sensor elements are projected on a smaller area (implemented in some DTP scanners), (ii) the resolution can increase independently of the scan dimensions (implemented in Vexcel VX 3000, and Leica/Helava DSW 200). 2. Electronic zoom Thereby, the optical resolution remains always the same but the signal is low-pass filtered and resampled in both (area CCDs) or one (linear CCDs) direction. In the latter case, the subsampling in scanning direction is accomplished by increasing the scanning speed by the same factor as the resolution decreases. This type of subsampling of linear CCDs leads to different treatment of horizontal and vertical direction and maybe also to certain image degradations, if the scanning movement is not increased by the proper amount. 3. Varying pixel size This is implemented in devices scanning one point at a time by use of apertures with different size and form. It is mainly used in microdensitometers and expensive drum scanners of large documents. 4. Microscanning This approach can be implemented by programmable area interline CCDs of standard resolution that permit a micro-movement of the sensor in order to increase the resolution (implemented in the Signum HIRES scanner). 5. Use of multiple sensors This is a rather rare case and is encountered in the Leica/Helava DSW 100 which has two area CCDs, a low and high resolution one.

11 6. Software zoom The scan is always performed in one resolution and the image is subsequently subsampled to lower resolutions using software. This can occur either after the scanning or during scanning if the scanner hardware is fast enough (latter case related to electronic zoom). Cases 3-5 are rare and/or expensive. Optical zoom using stable and precise optomechanical systems is faster than software zoom but also more expensive, and requires more careful calibration. Software zoom can be totally or partly implemented in hardware (e.g. in the latter case by subsampling in real-time the grey levels along a linear CCD). In certain cases it is advantageous to have images of different resolutions, e.g. image pyramids in photogrammetric workstations, for mensuration and display purposes. In such cases, there is no disadvantage in scanning at high resolutions and then subsampling using software. Electronic zoom is simple, fast, and does not require complicated calibration or expensive optomechanical parts, but requires accurate setting of the scanning speed and leads to a smearing of horizontal lines and different resolution in horizontal and vertical direction. The second task is scanning with different resolutions different parts (subimages) of one image. This may be useful, e.g. for measuring details like signalised points and fiducials with high resolution, while using the whole image in coarser resolution for a task like ortho-image or DTM generation. The problem here is the establishment of a correct geometric relation between the different subimages, i.e. all must refer to a common pixel coordinate system. Of course in any case the fiducials must also be scanned and refer to the same pixel coordinate system. To solve this problem the following approaches can be applied. 1. Use of accurate positioning mechanisms Thus, the top left of each subimage refers to the scanner coordinate system (implemented in the Zeiss/Intergraph PhotoScan 1, and Leica/Helava DSW 200 and DSW 100 scanners). 2. Scanning of patterns with known coordinates (common reference system) To avoid interference with the image the reference system, usually a réseau, is scanned separately from the image. The known coordinates of the crosses/dots form the common reference, whereby the spacing and the size of the crosses/dots should be chosen such that ambiguities are avoided and the crosses are measurable even for coarse resolutions (implemented in the Vexcel VX 3000 scanner). 3. Multiple scans The whole image could be scanned with high resolution and displayed but not saved on disk. If the scanner software supports measurement of pixel coordinates and cut-out of subimages (ROIs), then the ROIs that should be scanned with high resolution could be cutout and saved, as well as the pixel coordinates of the upper left corner of each ROI. Thus, all ROIs can be referenced to the scanner coordinates of the upper left corner of the whole image. Then, the image could be rescanned without changing any scanning parameters except the resolution (so, the scanner coordinates of the upper left corner remain the same), and the procedure repeated as often as it is required. Since the scanner, after each scan moves to a so called home (starting) position, the scanner must be able each time to position at the same upper left scanner coordinates (high repeatability). An alternative is to use some reference points (e.g. 3 known crosses in a L shape) to relate the starting position of each scan to a common reference system. 4. Software generated multiresolution images The image is scanned in the highest resolution that is required, and then an image pyramid is created, or the separate discrete resolution levels that are required. If only small parts of each pyramid level are needed, then the respective ROIs can be cut-out and the image discarded. By using the known geometric relations between the pyramid levels and the offsets of the ROIs all pixel coordinates can be referenced to a common pixel coordinate system. The first approach is the fastest, but also expensive. The second one is very flexible, but requires higher scanning (processing) time. It fits more to the notion of an interactive measuring device than a scanner for batch jobs. The third approach requires a certain software development by the user, and use of few reference points or that the scanner has high repeatability. This may be possible for scanners (especially scanners that can scan in one swath), which may otherwise exhibit positioning errors and nonuniform movement while scanning, i.e. the accuracy may be low but the repeatability high. A disadvantage is that if the scanner needs time consuming calibration, this must be repeated for each scan. The last approach is the cheapest, and easy to implement. Calibration is performed only once and no reference marks or high repeatability are needed. If the processing power is high enough, this approach can be faster than the second and the third one Photogrammetric functions Some photogrammetric scanners offer the possibility to perform certain photogrammetric tasks like measurement of fiducials or other reference points, and scanning in the photo coordinate systems. The measurement of the fiducials does not have to be a part of the scanner software, i.e. it can be performed later by photogrammetric software, if this information is not going to be used during the scanning. The claimed use is the exact positioning of each pixel in the photo coordinate system, and the possibility to scan in the photo coordinate system. The first requirement can be fulfilled by a geometric calibration and a fiducial

12 measurement/interior orientation after scanning. The second one can be performed either by aligning physically the sensor to the photo coordinate system (e.g. precisely rotating linear CCD of Zeiss/Intergraph PhotoScan 1) or by transforming the digitised data via software and putting the individual CCD frames together (e.g. in Leica/Helava DSW 200, DSW 100 and Vexcel VX 3000). The second approach can be also implemented after scanning. After performing the interior orientation, the image can be rotated and resampled. This can be performed fairly fast with the currently available processing power. In any case, the benefits of scanning in the photo coordinate system are small. It can only be used to help stereoscopic display by reducing the y-parallax. Since the rotation can account only for the k-rotation between images, no epipolar images can be created by this method. Thus, if epipolar images are needed for a subsequent image matching and DTM generation, this must be performed accurately in software, after the relative orientation. Thus, the benefits are small, while in the case of the Zeiss/Intergraph scanner the costs are high (the high precision servo-controlled spindles for rotating the linear CCD account for an essential part of the total scanner cost ; since few users will ever use this feature it should be offered to the customers as an option) Geometric and radiometric problems and tests Geometric and radiometric calibration procedures are usually applied by all vendors of photogrammetric and DTP scanners but in many cases they are incomplete, time consuming or not accurate enough. Photogrammetric scanners are well calibrated with respect to geometry but some of them exhibit significant radiometric problems like dark or light vertical stripes, other noise patterns, saturation of grey levels (especially in images with high contrast), and radiometric differences between neighbouring swaths scanned with linear CCDs. These problems indicate that radiometry is somewhat underestimated by the photogrammetric scanner vendors and the photogrammetric community in general. However, radiometry is very important since radiometric problems not only lead to poor visual results and make interpretation more difficult, but also influence the geometric accuracy of the measurements. In DTP scanners on the other hand, apart from having similar radiometric problems, geometric calibration is not implemented, or if it is, patterns and procedures of low geometric accuracy are used. Calibration and test procedures can and should also be applied by the user periodically. For such calibration procedures software and test patterns should ideally be supplied by the scanner vendors but this is unfortunately a rare case. In addition, the scanner vendors rarely provide the users with all relevant technical specifications of the scanner and with error specifications, e.g. tolerances for the RMS and maximum error that can occur in different cases. 3. SCANNER SPECIFICATIONS FOR PHOTO- GRAMMETRIC TASKS An optimal photogrammetric scanner should have the characteristics listed below. Some of them are more important than other ones. Characteristics, other than the ones mentioned here, are also acceptable as long as they do not lead to degradations of the scanner quality or high costs. Optimal characteristics of photogrammetric scanners include: Flatbed stage (high geometric accuracy, scanning of rigid documents). Optically butted linear CCDs. Many elements per CCD, butted precisely, 3-chip CCD desirable, low noise (dark current, fixed pattern, crosstalk and blooming), no defect pixels, uniform response, high maximum charge storage capacity, readout without multiplexing of the butted CCDs (i.e. single read-out register per CCD array), adaptive scanning for each colour channel to achieve colour balancing. An interesting technology is also Time Delay and Integration (TDI) scanning which permits a higher scanning speed or a higher SNR (such a technique is employed e.g. in Wehrli s RM 1 (Bethel, 1994), and the 10 x 15 cm format Polaroid CS 500 scanner). Note: single line CCDs could also be used, if a correct geometric and radiometric fit of all line swaths is implemented, but require more expensive positioning mechanisms and time-consuming calibrations. Area CCDs are attractive for interactive acquisition and revisiting of small image areas. Colour scanning preferably in one scan. Registration of colour channels with an accuracy corresponding to the positional accuracy of the scanner. Calibration (radiometric and if necessary geometric for each scanning stage for which calibration is required, i.e. for each scan and each colour channel or even for each partial scan, e.g. line swath or area tile), corrections implemented in hardware as much as possible, calibration software and test patterns provided, geometric errors constant over a long time period. Variable geometric resolution without change of the scan size, highest resolution of at least 1200 dpi. 10 to 12-bit quantisation with freely definable reduction to 8-bit, density range greater than 2.5D (preferably ca. 3.5D). Uniform, stable, diffused, white illumination, no heating of the scanner sensitive parts. The whole system should be designed such that the power of the illumination is the minimum possible. Avoidance of flare light particularly when scanning transparencies. Clean and stable power supply.